The Ripple Effect and the Linguistic Border in Belgium : A Country Divided?

A large literature has emerged, especially in the UK, that investigates regional convergence of house prices. Many authors have found regional house prices to be converging in the long-run and exhibit a distinct spatial pattern over time, which has become known as the ripple effect hypothesis. In this paper we examine the validity of the ripple effect hypothesis for Belgium and are particularly interested in the role of the linguistic border in the spatial and temporal propagation of shocks in a dynamic system. We extend the model that was recently proposed by Holly et al. (2011) to cope with the unique federal structure of Belgium and use data at the level of the judicial districts (N = 20) for an extensive time period (1973Q1-2011Q3, T = 155). We show that the linguistic border plays an ambiguous role. The results indicate that almost all regional house prices are converging in the long-run, which implies that regional markets in Belgium are integrated. We furthermore show that house prices in regions which are located along the north-south axis in Belgium, which constitutes the economic spine of Belgium, converge more quickly with respect to house prices in the dominant region, Antwerp. This result suggests that the linguistic border plays no significant role in the house price diffusion process. After this initial error correction mechanism, however, the convergence process follows a distinct linguistic pattern (east-west axis) where regions converge only with respect to neighboring regions that are located within the same linguistic region. Moreover, short-run spatial spill-overs are significant for nearly all neighboring regions that lie within the same linguistic area, but nearly nonexistent for neighboring regions across the linguistic border. Finally, we provide evidence for the ripple effect hypothesis in Belgium