Latent grouped structures in panel data: a review

Latent group structures in panel data models are a new and powerful approach to deal with unobserved heterogeneity in a parsimonious way. This review, with a special focus on grouped structure in unobservable traits, first analyzes the limits and opportunities of Bonhomme and Manresa (2015a)’s Grouped Fixed Effects (GFE) estimator, also discussing the literature it contributed to create. A rich selection of models enhancing clustered heterogeneity at a slope level, starting from Su et al. (2016a), is then presented. A short section investigates how the applied literature has employed in practice the GFE. Finally, the GFE of Bonhomme et al. (2022) is presented in detail together with its limits and advantages.