Exploring spatial convergence of Maghreb regional areas: An application of a Markov chains approach

This paper explores the real convergence process of 360 regional areas in Maghreb countries using a non-parametric approach based on classical and spatial Markov chains. This approach investigates the long-term spatial associations between regional units. The advantage of this method is that is does not require the use of control variables which are very often non available at a disaggregated geographical level. We show that there is a trend of regional convergence of GDP per capita in Mena regions. However, this process is very slow and there is an important spatial clustering movement. We also show that although almost 75% of the regional areas seem to converge towards a higher GDP per capita level in the stationary state, 25% are trapped in low development trends. These results suggest that the economic performance of a given region strongly depends of that of its neighbor regions. This in turn implies reconsidering regional development policies within these countries.