A regional unemployment model simultaneously accounting for serial dynamics, spatial dependence and common factors

Abstract

Regional unemployment rates tend to be strongly correlated over time, parallel the national unemployment rate, and be correlated across space. We address these key stylized facts by linking different strands of literature into a unified methodology to investigate regional unemployment disparities. This methodology simultaneously accounts for serial dynamics, spatial dependence and common factors, also known as weak and strong cross-sectional dependence. We apply this approach using provincial level data for the Netherlands. The substantial and persistent division between high and low unemployment clusters makes it an interesting case, and data availability since the early 1970s enables a comparison between prior periods of downturn and recovery to the recent economic crisis. It is found that approaches that do not simultaneously account for serial dynamics, spatial dependence and common factors, or that ignore one of these issues, may lead to biased inference.