Voter Fatigue in Straightforward Decisions A Natural Experiment from a Recall Election

Voter Fatigue in Straightforward Decisions: A Natural Experiment from a Recall Election Extended Abstract While in most countries all citizens have the right to vote, a substantial body of literature studies the heterogeneity in the degree to which voters are actually represented by their votes. Among other things, this heterogeneity arises from the quantity and quality of information consumed (Lee et al 2014). In addition, several electoral processes require voters to cast votes on many issues, potentially leading to choice overload. Augenblick and Nicholson (2016) exploit a natural experiment to show that voter fatigue can appear in the setting of elections in California, where voters had to choose among a wide variety of issues. In this paper, we show that voter fatigue can appear also in the setting of relatively straightforward decisions, where voters had to choose whether to recall or not a mayor and her councilpeople. Setting and Research Design The setting is the 2013 recall election of the mayor of Lima, Peru’s capital. The recall election was for mayor Susana Villaran and her council. The council is formed by 39 people, of whom 21 were from the mayor’s party and 18 were from other parties. The ballot had all 40 names split in two columns. Next to each name, people had to vote for or against the recall by marking “yes” or “no” in the ballot. The first name on the list is the mayor’s, followed by the councilpeople from her party. After them came the councilpeople from other parties, ordered by the share of votes each party received in the municipal election. In this case, it so happened that the mayor’s party had the entire first column and the first two positions in the second column. The next party followed from the third position in the second column onwards. The ballot does not indicate the party to which each person is affiliated. As shown in Figure 1, it is not possible from the ballot to identify the party to which each person belongs. Figure 1 Ballot from the Lima 2013 mayor recall election Note: colored rectangles were added by the authors for clarity of exposition. They were not part of the ballot. In municipal elections, citizens cast their vote for the mayor only, not for councilpeople. Each party running for the mayor’s seat puts forward a list of councilpeople and the seats are awarded based on the share of votes for the mayor. Councilpeople are basically unknown to the public. In a setting were voters are not asked to vote for individual councilpeople in the election but are asked to do so in the recall, some possible strategies by voters could be: “recall the mayor only”; “recall the mayor and all her party’s councilpeople”; “recall everyone”; or “recall everyone but those from the party I voted for”. If a voter follows the strategy to vote for (or against) the mayor and all her councilpeople and leave the rest blank, he or she may miss the last two due to voter fatigue. It is difficult to argue that “recall the mayor and all but two of her council people” was a strategy that voters actively chose. Our measure of fatigue is the share of null and blank votes for each councilperson. We show that this share increases discontinuously when voters have to switch to vote for the councilpeople in the second column, despite being from the same party as the first column. Data – Our main data source is voting-booth data from the 35,830 voting booths in Metropolitan Lima, provided by the National Bureau of Electoral Processes (ONPE, in Spanish). Results Our main results are presented in Figure 1. The share of blank and null votes increased significantly between the first and second column, by 2.4 percentage points. Since the councilpeople immediately to the right of the discontinuity are from the same party as those in the left, we argue this jump is a result of voter fatigue. Figure 3 shows a negative association between average district income and voter fatigue. In the poorest districts voter fatigue, measured by the discontinuity in blank and null votes at the threshold, is 38% larger than in the richest districts (2.9 vs 2.1 percentage points). This finding implies that, despite on paper each citizen has the same right to vote, voters from poorer districts are de facto less represented by their votes than voters from richer districts. Second column in ballot 10 30 .12 % of blank or null votes First column in ballot Porcentaje de votos nulos o en blanco .14 .16 .18 .2 Figure 2 Share of blank or null votes by ballot position 0 20 Posición en la balota 40 Ballot Position Note: This figure shows that the share of blank or null votes increased significantly between the 20th and the 21st name in the ballot. The former is the last council member to appear in the first column, and the latter is the first council member to appear in the second column of the ballot. We attribute the discontinuity to voter fatigue. Figure 3 Voter Fatigue and District Mean Income Notes: The vertical axis is the district’s average monthly income in Soles (S/ 1,000 = US$ 270). The horizontal axis is average discontinuity in the district. The figure suggests that voter fatigue is stronger in lower-income districts.