Campaigning in Mixed Electoral Systems: A Contest Theory Approach

• In Event: Models of Electoral Competition

Abstract

In this study, I develop a model of political competition: multiple parties engage in electoral competition to win a limited number of seats in the parliament. More specifically, I study political candidates’ incentives or campaign behaviors in mixed electoral systems: how do candidates allocate resources towards personalized versus party-centric campaigning? I use a contest theory model to derive theoretical predictions and test those predictions using an online experiment.

Electoral formulas can affect the incentives or campaign behaviours of political candidates. For example, with the First-Past-The-Post (FPTP) system, votes are counted at the individual level in local districts, and political candidates may focus more on improving local projects or running more personalized campaigning to attract more votes. With a Proportional Representation (PR) system, the votes are pooled at the party level and candidates may focus more on national projects and campaigns with more party-centric messages. The candidates’ campaign strategies become less clear with mixed election systems.

A mixed electoral system is in general a combination of FPTP and PR and is adopted in countries like Germany, Japan, South Korea, etc. Candidates are in general allowed to both run as a candidate at the local district and register on the party list. Dual candidates will face a trade-off when they have a limited budget: how many resources will they spend on local projects or personalized campaigning to help them succeed at the district level? How many resources will they put towards national projects or party-level campaigning to help their parties gain more seats? This question has important implications for both parties and institution designers.

I model this trade-off using a contest theory model. More specifically, candidates need to choose how much resources they want to put into ‘district-level battle’ and ‘party-level battle’. At the district-level battle, candidates from both parties form a pairwise match and whoever wins gets a seat in the parliament. The rest of the seats will be determined at the party-level battle, the seats will be distributed to parties with proportional representation. Then the parties distribute the winning seats to their party members. The winners of those battles are determined using a Tullock Lottery success function. Candidates only care about winning a seat in the parliament.

Consider the case in which the party-level seat is distributed inside the party according to a pre-determined list of candidates, or a closed list. I find that the lower-ranked candidates choose their battles: they spend all their resources towards the district-level battle. The intuition is such that they have very little chance of getting the party-level seat with the low rank, so they spend all they have to win at the district level. The high-ranked candidates will split resources across both battlefields. The allocation of resources also depends on the pairings at the district level. When the high-ranked needs to compete with low-ranked candidates who are a lot more aggressive at the district level, the high-ranked move more resources towards the party.

When the party-level seat is distributed according to a merit rule or an open list, those who spend more resources towards the party are also more likely to get the seat. In this case, all candidates spread resources across two battlefields. Comparing total party contributions between a closed list and an open list, total contributions with an open list can never be smaller than under a closed list. This is because that the intra-party competition element reduces the free-riding incentives with a closed list.

I also consider a few extensions to the benchmark model. Firstly, spending resources on one battlefield can have positive spillover effects on the other: a charismatic personal campaign may help leave a better impression of the party. Moreover, candidates may attach special meanings to win through a certain battlefield: candidates may value winning through the district because of the close links to local communities. Additionally, within a party, some candidates may receive a greater campaign budget than others.

To understand this trade-off better, I also plan to run an online experiment based on the theoretical framework. In the experiment, individuals from different teams compete for a number of indivisible prizes. Participants will receive a certain amount of endowment at the start of each round and decide how much to invest in their ‘individual box’ versus the ‘team box’. Different treatment groups differ in the prize sharing within the winning team and the pairings at the individual competitions. This experiment will use these treatment groups to test the effect of different prize-sharing rules on the allocation of resources; how different pairings at the individual competitions affect the allocation when the winning team distributes the team-level prize according to a list rule; and how the rank of participants affects their allocation decisions.

Author

• Yixuan Shi, Max Planck Institute for Tax Law and Public Finance