Campaign Finance Levels as Coordinating Signals in Three-Way, Experimental Elections

To the extent that election campaigns are simply advertising campaigns, they should not contribute directly to social welfare. Yet election campaigns are often very costly. Why do such costly campaigns arise as the norm? If they do not benefit society, should campaign expenditures be limited? However, costly election campaigns can benefit society indirectly by providing voters with a signal about how much others support each candidate. With this signal, voters may be able to avoid coordination failures that lead to the Condorcet loser paradox. Here, following Myerson and Weber (American Political Science Review, 1993), we model this situation using a simple game theory argument. Then, we study the effectiveness of campaign finance levels as signals in laboratory elections. We find that finance levels are quite effective in coordinating the majority and defeating Condorcet losers. We also find that campaign finance levels are efficient in that: (1) the total benefits to coordination exceed the costs of campaigns, (2) the last typical incremental contribution increases the chances of coordination sufficiently to exceed the cost of the contribution and (3) the next typical incremental contribution would not have increased the chances of coordination sufficiently to exceed the cost. We also find that voters are not subject to a "sunk cost" effect that may influence them to vote for a losing candidate simply because they had contributed to that candidate in the past. Instead, they seem perfectly willing to "cross over" to another candidate when coordination requires it. Finally, we show briefly how this methodology can be extended to study other election systems. In particular, we discuss the relative costs of campaigns under approval voting