We analyze an infinite horizon model of bargaining over a set of policies, in which both players receive utility in every period, and preferences evolve over time. In each period one player makes a take-it-or-leave-it offer. If the other players accepts the offer, the new policy is implemented. Otherwise, the previous period policy remains in place. In the subsequent period, the previously implemented policy becomes the default option, and the process repeats itself. The endogeneity of status quo affects the behavior of the players: in each period they trade off the current utility with the future bargaining power determined by the implemented policy. As a result, the policies implemented in each period may not be Pareto efficient. We examine how the efficiency and the inertia of policy making is affected by this dynamic linkage.

Existing envelope theorems apply to fixed choice sets or to convex maximization problems (Milgrom and Segal 2002). We derive envelope theorems for parametrized choice sets without imposing any convex or topological structure on the choice sets. We show that the traditional envelope theorem formula hold at any point where the generalized Lagrange multipliers and the constraint are sufficiently smooth in the parameter. We provide conditions under which the value function is differentiable or absolutely continuous. We apply these theorems to mechanism design problems.

We consider a simple model of policy coordination among heterogeneous countries and examine the consequences of national elections on the degree of international coordination. We show that even perfectly open and competitive national electoral process can bias the international policy outcome compared to the case in which benevolent national policy makers maximize national welfare. Under direct democracy, if the distribution of preferences is skewed towards autarkic preferences, the irrationality of national majority preferences exerts an additional bias towards less policy coordination. Under representative democracy, voters elect representatives with more extreme preferences than themselves, more so the more extreme the representative of the other country. Finally, a more cooperative policy process distorts national elections and results in less policy coordination.

We compare centralization versus decentralization in a federation in which members have different preferences but policy heterogeneity is inherently costly. Unlike the existing literature on fiscal competition and environmental spillovers, externalities are driven by the differences and incompatibilities between local policies. Contrary to the common wisdom, decentralization becomes optimal as coordination costs increase. When externalities are reciprocal and sufficiently elastic, decentralization dominates centralization whatever the magnitude of externalities and the heterogeneity of preferences. In the case of discontinuous network effects, uniformization never Pareto dominates decentralization.

We revisit the common assumption of most model of federalism according to which centralized governments are compelled to implement uniform policies. We consider a federal system which exhibits cross-border externalities along with heterogeneous local preferences. We assume that the latter are not known by the central administration. We show that strategyproof centralized mechanism cannot unambiguously improve on the traditional decentralized equilibrium. Moreover, neutral and group strategyproof mechanisms are necessarily uniform.

We generalize a result due to Karakosta and Kotsogiannis (2006) who showed in a two country model with imperfect competition, linear technologies and product market taxation that the horizontal externality effect dominate the vertical one (i.e. that local taxes are too low) if and only if at the multi layer tax equilibrium, the marginal welfare of the local public goods is higher than the one of the federal public good. We show in a simple proof that this result generalize to any number of country, firms, any technologies and any kind of taxation.

We consider a federation in which heterogeneous member states vote on a common discretionary interval in which they can chose freely their own policy (tax, law, regulation). This interval restricts the residual right of control of the states in order to balance the coordination failure that arises under complete decentralisation with the rigidity of unitarian centralisation. We consider a voting mechanism in which the states first vote on the size of the discretionary interval (the degree of centralization), then on its location (the federal directive) and finally each member chooses its policy within the federal bounds. We show that in equilibrium, the location of the interval is negatively sensitive to the preferences of states with non median preferences. Hence, our voting mechanism endogenizes the usual preferences matching problem of centralization. Moreover, whatever the severity of cross border externalities, at least a majority of states will be constrained by the federal bounds.