Title: Dynamic bayesian models for panel data with dynamic skewness
Authors: Hasan Uri - The University of Sheffield (United Kingdom) [presenting]
Abstract: Time series models often rely on symmetry of the innovation process to carry out estimation. There are several alternatives in the literature extending some of the most commonly used models to include skewness. I propose a further extension, allowing for the skewness in the innovations to vary with time. This can be useful in, for instance, defining new measures of convergence in economics and finance. I use the reciprocal skewing mechanism and propose an autoregressive gamma process to model the path of skewness. This is embedded in an linear AR(1) model for panel data from a Bayesian perspective, so I provide a hierarchical prior structure suitable for most applications envisaged and discuss sensitivity to prior parameters. The methodology is illustrated using data on EU economic growth.