The traditional approach to modeling political parties' behavior, based upon the contribution of Anthony Downs (1957), assumes that the parties' unique objective is to win elections: thus, they maximize their popularity. The crucial implication of this assumption for a two-party system is that if the two parties have the same information about voters' preferences, full convergence of policies results from electoral competition. This is the crucial implication of the median voter theorem. ' More generally, it may be argued that different parties are differently because they represent different constituencies. Parties may not care only about winning elections per se, but also about the quality of the policies resulting from an election. In this case the candidates of the two parties view winning an election not only as a goal per se, but also as a means of implementing a better policy for their respective constituencies. This paper shows that electoral competitions imply dynamic inconsistency if the voters are modeled as rational and forwardlooking agents and parties do not care exclusively about being elected, but also about which policy to implement, once elected. The dynamic inconsistency arises as follows: the parties have an incentive to announce convergent platforms to increase their chances of election. However, if the elected party is not committed to its electoral platform, it has an incentive to follow its most preferred policy rather than the policy announced in its platform. If voters are rational, they account for this incentive. Thus, in general, in a one-shot electoral game the only timeconsistent equilibrium is one in which no convergence is possible, the two parties follow their most preferred policies, and the voters rationally expect this outcome. Full convergence of parties' platforms results only as a limiting case when the parties are completely indifferent with respect to the quality of the policies resulting from the election. Thus, these results differ from the existing literature on ideologically motivated politicians (for instance, Donald Wittman, 1977, 1983; Randall Calvert, 1985), which implicitly assumes the possibility of binding commitments to electoral platforms. Complete or partial policy convergence can be the outcome of political competition if the interaction between the parties and the voters is modeled as an infinitely repeated game. In fact, if the candidates have concave objective functions, the welfare-maximizing policy rule implies a complete convergence of parties' policies. However, this cooperative, and agreed-upon policy, may or may not be sustainable as a subgame-perfect equilibrium depending on parameter values; in particular it depends on the discount rates of the two parties, the degree of polarization of their preferences, and the relative popu*Graduate School of Industrial Administration, Carnegie Mellon University, Pittsburgh, PA 15213, and National Bureau of Economic Research, Cambridge MA, 02138. This paper is based upon a chapter of my unpublished doctoral dissertation at Harvard University. I am greatly indebted to Jeffrey Sachs for directing my attention toward these issues and for many conversations. I also wish to thank Andrew Abel, Dilip Abreu, Olivier Blanchard, Ramon Caminal, Andrew Caplin, Alex Cukierman, Morris Fiorina, Benjamin Friedman, Herschel Grossman, Howard Rosenthal, and the referees for very useful comments. The responsibilitv. Af anv n mtnkieis Af cAlrirp nfnlv mine The result of policy convergence in a two-party system is more general than the median voter theorem. For discussions of convergence results not at the median, see John Ledyard, 1984; Peter Coughlin, 1984; Coughlin and Shmuel Nitzan, 1981; Melvin Hinich, 1977. For earlier work on spatial competition see Richard McKelvey, 1975; Hinich, Ledyard, and Peter Ordeshook, 1972, 1973, and the references quoted therein. The present paper focuses on the result of convergence rather than on the median voter theorem per se.

Critics of foreign aid programs argue that these funds often support corrupt governments and inefficient bureaucracies. Supporters argue that foreign aid can be used to reward good governments. This paper documents that there is no evidence that less corrupt governments receive more foreign aid. On the contrary, according to some measures of corruption, more corrupt governments receive more aid. Also, we could not find any evidence that an increase in foreign aid reduces corruption.

In all industrial countries, fiscal policy is increasingly about redistribution. In this paper, we study redistribution across different types of agents in a world characterized by the presence of labor unions and distortionary taxation. We show that an increase in transfers financed by distortionary taxation has nonlinear effects on unit labor costs relative to the other countries, depending on the degree of centralization of the wage-setting process in the labor market. We find considerable empirical support for the model in a sample of 14 OECD countries.

A widely held view suggests that politically independent central banks bring about relatively low and stable inflation rates.' A more debated question is whether one has to pay for this good outcome with more real instability. In his seminal contribution, Kenneth Rogoff (1985) suggests that an independent and inflation-averse central bank reduces average inflation but, as a result, increases output variability; the conservative central banker reduces the inflation bias, due to the time-inconsistency problem, but stabilizes less. However, Alesina and Summers (1993) do not find that, at least within the OECD countries, more independent central banks are associated with more variability of growth or unemployment. Thus, they conclude that independent central banks bring about low inflation at no apparent real costs. The point of this paper is to provide theoretical underpinnings to this finding, which is in contrast to Rogoff (1985).2 The basic idea is that one can isolate two sources of output variability. One is the economic variability induced by standard exogenous shocks that monetary policy is supposed to stabilize, for instance, money demand shocks or supply shocks. The second source of variability is or, more generally, policy-induced. This is the variability introduced in the system by the uncertainty about the future course of policy. For instance, Alesina (1987) studies the effect of uncertain electoral outcomes in a model where the two contending parties have different preferences over inflation and unemployment. An inflation-averse, independent central banker does not stabilize as much the economic variability, in order to keep inflation low and stable. This is Rogoff's point. However, by insulating monetary policy from political pressures, an independent central bank can reduce the variability. The overall effect of independence on output variability is, thus, ambiguous. This result is consistent, at least prima facie, with the evidence in Alesina and Summers (1993) on the lack of correlation between centralbank independence and output variability. In fact, it is possible that when the politically induced output variability is predominant, a more independent central bank reduces average inflation and the variance of output.

When a stabilization has significant distributional implications (e.g., tax increases to eliminate a large budget deficit), socioeconomic groups may attempt to shift the burden of stabilization onto other groups. The process leading to stabilization becomes a "war of attrition," each group attempting to wait the others out and stabilization occurring only when one group concedes and bears a disproportionate share of the burden. We solve for the expected time of stabilization in a model of "rational" delay and relate it to several political and economic variables. We motivate this approach and its results by comparison to historical and current episodes.

This paper analyzes a model in which a group of rational individuals votes over the composition and time profile of public spending. All voters agree that a balanced budget is ex ante optimal. However, if there is disagreement between current and future majorities, a balanced budget is not a political equilibrium under majority rule. Under certain conditions a majority of the voters favors a budget deficit, and the equilibrium deficit is larger the greater is the polarization among voters.