Citizen-candidates, lobbies, and strategic campaigning (Q2385116)

The innovation of the paper is to provide a model in which agents can incur a cost to support a policy which is not their most preferred policy. The author is interested in studying the phenomenon of ``vote stealing.'' Without ego rents, a three-entrant equilibrium in Osborne and Slivinski's model does not exist. This result reframes in the following way. In a modified version of the Osborne and Slivinski model with no ego rents and for which citizen-candidates are allowed to campaign on any policy at all (incurring a cost), there cannot exist a three-entrant equilibrium in which each entrant campaigns on their ideal point. This fundamental observation provides the basic intuition underlying the general results of author. The author studies the standard single-dimensional spatial model of politics. The assumptions are: lobbies have preferences over policies, and money that are additively separable. A lobby's utility function over a policy \(x\) and money spent \(m\) can be written as \(U(x,m)=u(x)-\psi(m)\), where u is concave. While most lobbies have an ideal policy, they need not pledge money to their ideal policy (or to any policy at all). It is interested in studying the conditions under which lobbies pledge money to policies they do not like. The designed model is a one-shot game, in which lobbies simultaneously pledge money to different policies. There is a second (implicit) stage where the election is conducted among the entrants, this stage is entirely non-strategic (so that voters always vote for their most preferred entrant) and is therefore not part of the game. The main finding in this paper is that, in fact, the converse statement is true. If there are at least three entrants, then at least one lobby finances an entrant it does not like. Therefore, ``truthful financing'' obtains if and only if the number of entrants in the election is at most two. For all \(l\in{L}\), define a utility function representing preferences, \(U_{l}:S\rightarrow R\), as \[ U_{l}(S)=\frac{\sum_{x\in{W(s)}}u_{l}(x)}{|W(s)|}-\psi_{l}(m_{l}). \] Thus \(U_{l}\) is the expected utility of a lobby \(l\) when the set of winners is \(W(s)\), and when the amount of money spent by \(l\) is \(m_{l}\). To recap, each lobby's strategy space is \(S_{l}=X\times[0,\omega_{l}]\), and each agent's payoff function is given by \(S_{l}:S\rightarrow R\) as specified above. The author studies the pure-strategy Nash equilibria of the election game: \((L,S,\{U_{l}\}_{l\in{L}},\mu)\). The first theorem provides a simple set of necessary conditions that a strategy profile must satisfy if it is a Nash equilibrium of the election game with only one entrant.The next theorem provides a set of necessary conditions that any Nash equilibrium of the election game with only two entrants must satisfy. As in the preceding theorem, the conditions are also sufficient if no lobby can finance a candidate on its own. The model is reminiscent of the citizen-candidate models of both Osborne and Slivinski, and Besley and Coate. But the model is closer to that of Osborne and Slivinski, who also assume a sincere-voting stage; whereas Besley and Coate assume that voters are themselves strategic actors. Both models assume that entrants must choose their favorite policy on which to campaign.