The globalization process has led to a growing interdependence of fiscal policies affecting, inter alia, the composition of government expenditures. Sanz and Velazquez (2004) show that there was convergence in the structure of government expenditures by functions in the Organisation for Economic Co-operation and Development (OECD) during the period 1970-97. Convergent forces may be expected to be even more intensive in the case of EU member states involved in the integration process. First, interdependencies of economic policies are greater with respect to fiscal policy. Second, these economies may have more similar productive structures, which under the assumption of growth-maximizing governments could lead to similar public expenditure compositions. Therefore, a greater convergence process might be expected in the composition of government expenditures by functions in the EU than in the OECD.
In this study, the authors assess whether this harmonization process has been greater in the case of the EU member states, so that convergence may be partially attributed to the European integration process. For this purpose, section II examines the convergent forces leading to similarity in the composition of government expenditures by functions. Furthermore, it analyzes why higher convergence should be expected for EU Member States than for the OECD. In section III, an annual dissimilarity index is constructed to explore the differences between the structures of government expenditure within the EU using the non-EU countries of the OECD as a control. Section IV adapts the usual indicators of the income convergence literature to examine the composition of public expenditure in the EU and OECD. Moreover, the authors investigate whether this convergence is conditional rather than absolute and whether opportunities for future convergence remain. Section V sets out the main conclusions.
II. CONVERGENT FORCES OF THE COMPOSITION OF GOVERNMENT EXPENDITURES
Following Barro (1990), Devarajan et al. (1996) elaborate an endogenous growth model, where a representative infinitely lived agent chooses consumption and capital to maximize a utility function subject to a budget constraint. A production function for a single good has three inputs: private capital and two components of government spending, productive and nonproductive expenditures. (1) In this way, these authors obtain an expression for steady-state growth in terms of the shares of each component in total government expenditures. Indeed, Davoodi and Zou (1998) and Xie et al. (1999) use this model, though they differentiate government expenditure by levels (federal, state, and local) instead of the functional character of expenditure. Thus, assuming a Cobb-Douglas production function, these authors conclude that the growth-maximizing shares of each component in total government expenditures are:
(1)[[phi]*.sub.c] = [[beta].sub.c]/[[[summation].sub.c=1.sup.3][[beta].sub.c]] [3.summation over (c=1)][[phi]*.sub.c] = 1 c = 1,2,3.
Where [[beta].sub.c] is the elasticity of growth with respect to component c of public expenditure, c is one of the three levels of administration considered and [[phi]*.sub.c] is the optimal share of spending component c in total public expenditure. Therefore, there is an optimal composition of government expenditures in which the share of each component equals its relative growth elasticity. This model can be extended to N components of government expenditure and applied to any composition of government expenditures: by economic type, by levels of government, or by functions. Thus, if governments maximize growth, and output elasticities with respect to each component ([[beta].sub.c]) are similar across countries, one should expect convergence toward similar compositions of government expenditures.
Note that the authors assume that governments maximize economic growth as in Devarajan et al. …