AN ENDOGENOUS POLICY MODEL OF HIERARCHICAL GOVERNMENT 1 Isidoro Mazza * and Frans van Winden ** * University of Catania, Department of Economics and Quantitative Methods, c.so Italia 55, 95129 Catania, Italy. E-mail: imazza@mbox.unict.it ** University of Amsterdam, CREED and Tinbergen Institute. Address: Faculty of Economics and Econometrics, Roetersstraat 11, 1018 WB Amsterdam, The Netherlands. E-mail: F.A.A.M.vanWinden@uva.nl ABSTRACT Endogenous policy models usually neglect that government policies are frequently the result of decisions taken by different agents, each enjoying some degree of autonomy. In this paper, policies are the outcome of the choices made by two agents within a hierarchy: a legislator decides on the budget to be spent by a bureaucrat. Each policymaker can be lobbied by one or two interest groups. Our results, concerning the effects of multi-tier lobbying and legislatorial oversight, substantially qualify conventional wisdom related to one-tier lobbying. The potential advantages of a double opportunity to influence decisionmaking are confronted with the disadvantages of extra lobbying expenditure and separation of powers. In particular, the reaction of the legislator to lobbying at the bureaucratic tier may make lobbying wasteful even when there is no competition from other lobbies. In case of competition, the ability of a group to prevail depends on the relative access at both tiers. The analysis focuses on the effects of changes of influence across stages. Furthermore, competition for influence at the bureaucratic tier may work as a substitute for legislatorial oversight. Extensions of the model indicate its usefulness for the analysis of decisionmaking in other multilevel governance structures, like federations or firms. (JEL : D72, D73, D78, H39, H77). Keywords: Multi-tier lobbying, endogenous policymaking, hierarchy, multilevel government, bureaucracy, fiscal federalism. 1 Earlier versions of this paper circulated under the title “Two-tier lobbying”. For helpful comments and suggestions we wish to thank Chong-Bum An, Sanjeev Goyal, Kazuyuki Nakamura, Randolph Sloof, Alexander Tabarrok. The usual disclaimer applies.