Spatio-temporal analysis of regional systems : a multiregional spatial vector autoregressive model for Spain

This paper contributes to the recent literature in spatial econometrics that focuses on space-time data modeling implementing a multi-location time-series statistical framework to analyze a regional system. Therefore, taking as a point of departure the Global Vector Autoregression approach introduced in Pesaran et al. (2004), a multiregional spatial vector autoregressive model (MultiREG-SpVAR) is formulated and then applied to study the spatio-temporal transmission of macroeconomic shocks across the regions in Spain. The empirical application analyzes the extent to which a region's economic output growth is influenced by the growth of its neighbors (push-in or inward growth effect), and also investigates the relevance of spillovers derived from temporary region specific output growth shocks (push-out or outward growth effect). Our results identify some regions that perform as "growth generating" within the Spanish regional system since growth shocks from these regions spill over to a large number of regions of the country, playing a key role in transmitting regional business cycles. The policy implications of our results suggest that national and/or regional governments should stimulate economic activity in these leading regions in order to favor the economic recovery process of the whole Spanish economy.