Regionalization, public spending and growth: a stylized model dealing with ‘predatory states’ Elsa Duret a , Bruno Ventelou b, * ,1 a CERDI-IDREC, Universite ´ d’Auvergne, 65 Bd Franc ¸ois Mitterrand, 63 000 Clermont-Ferrand, France b INSERM U 379/ORS PACA, 23 rue Stanislas Torrents, 13006 Marseille, France Accepted 24 February 2004 Abstract In this paper, a micro-founded model analyzing the effects of ‘regionalization’ on economic activity is developed. It shows that the spatial division of public competencies can have an impact on the growth rate via the efficiency of governmental choices: initially advantageous for weak levels, decentralization (/reduction of regional size) becomes limited due to the risk of underestimation of the real profitability of public expenditure by local governments (non-internalized cross-border effects). In accordance with the theory, a transversal estimation for a sample of 51 countries for the 1990s establishes a ‘bell-shaped’ relation between indicators of regionalization and the quality of governance. D 2004 Elsevier B.V. All rights reserved. JEL classifications: D7; H7; R1 Keywords: Intergovernmental relations; Spatial economics; Public finance and corruption 1. Introduction The territorial organization of the State is a striking element of the complementarities installed between the private logics of accumulation and the public logics of allocation, regulation and redistribution. The question is how, and to what extent, the level of decentralization really favors growth? In order to answer this question, we suggest a theoretical model which analyses the effects of regionalization on the ‘predatory behavior’ of governments, showing a bell-shaped relation between these two variables: initially 0264-9993/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.econmod.2004.02.002 * Corresponding author. E-mail address: ventelou@marseille.inserm.fr (B. Ventelou). 1 Research fellow at OFCE-National Foundation for Political Sciences. www.elsevier.com/locate/econbase Economic Modelling 21 (2004) 1039 – 1050