Three Essays in Political Economy

Both economist and political scientist agree that the composition of the
constituency is an important driver of policy outcomes. The existing literature
has tried to find explanations why people turn out to vote. While we still lack
an adequate theory of political participation in large elections, there is a broad
consensus that rational turnout behavior comprises a trade-off between costs and
benefits of individual actors. The seminal work of Downs (1957) was the first to
stress the so called "paradox of (non-) voting". In large elections, individual's
probability to change the outcome is vanishingly small. Therefore, citizens
should abstain even in the presence of only minor positive costs. Yet, the theoretical
predictions of zero turnout are contradicted by positive rates of participation
in all Western countries. Recent approaches to rational voting turnout can be
broadly divided into three categories. The first strand endogenizes whether a
voter is pivotal (Ledyard, 1984; Palfrey and Rosenthal, 1983, 1985; Levine and
Palfrey, 2007). The second argues that information asymmetries dissuade people
from voting (Feddersen and Pesendorfer, 1996b, 1999; Battaglini and Morton,
2008). The third integrates the civic duty component, originally formulated by
Riker and Ordeshook (1968), into a structural model of group voting (Coate and
Conlin, 2004; Feddersen and Sandroni, 2006).

This thesis aims to make a contribution to the turnout literature by using the
system of direct democracy in Switzerland in combination with the country's
decentralized electoral legislation that generates considerable variation in the
costs to voting. In a nutshell, the second chapter examines the long-term effects
of a severely sanctioned voting norm in the canton of Vaud. The compulsory
voting norm had been in place for more than 25 years before being abolished in
1949. Individuals in Vaud were confronted with the same choices as their fellow
citizens in the rest of the country where compulsory voting was not introduced.
The third chapter explores whether cost shocks have an effect on the turnout
decision of voters. I use data from the post-vote survey (FORS 2011), which I
combine with a dataset on local rain data as well as information on the stepwise
introduction of postal voting in Swiss cantons. Both variables generate large
variations in the costs of voting. The fourth chapter focuses on the mobilization
power of referendums. Specifically, I classify proposals according to their a
priori mobilization power and test whether spillovers from concurrent proposals
change voting outcomes of less mobilizing proposals. The analysis explores
whether large spillovers increase the degree of status quo bias and government
support.

The empirical work on political participation has revealed that the turnout
decision depends on a number of factors. While social pressure (Gerber and
Green, 2000; Gerber, Green, and Larimer, 2008; Funk, 2010) and altruism
(Fowler, 2006) have been considered as important determinants of turnout for a
long time, recent contributions stress the role of genes (Fowler and Dawes, 2008)
and information asymmetries (Feddersen and Pesendorfer, 1996b; Battaglini and
Morton, 2008). However, costs and benefits are at the core of the rational choice
approach. Recent models of group voting (Coate and Conlin, 2004; Feddersen
and Sandroni, 2006) as well as models multidimensional voting decision models
(Krishna and Morgan, 2011a) and pivotal-voting models (Ledyard, 1984; Palfrey
and Rosenthal, 1983, 1985) base their theory on the assumption that voters
have heterogenous voting costs. As a consequence, rational voters use cutpoint
strategies which refers to the fact that there is exactly one voter who is indifferent
between his voting costs and the benefits from going to the polls. All individuals
with voting costs below this cutpoint value go to the polls, those with higher
voting costs abstain.

Chapter 2, which is joint work with Dominik Hangartner and Michael
Bechtel, focuses on the long-term and spillover effects of compulsory voting.
The chapters examines norm internalization in the context of a severely sanctioned
and long-standing compulsory voting law in the Swiss canton of Vaud.
Social norms such as fairness, conditional cooperation, or inequality aversion are
prominent explanations for the empirical puzzle that individuals often provide
public goods at levels that exceed those predicted by orthodox theory (Fischbacher
and Gächter, 2010; Rege and Telle, 2004; Fehr and Fischbacher, 2004;
Fehr and Gächter, 2001; Fehr and Schmidt, 1999; Lindbeck, 1997; Coleman,
1990; Hollaender, 1990; Ullmann-Margalit, 1977). The study of the evolution of
social norms has central to all social sciences, but most of the existing literature
focuses primarily on the evolution of norms in the absence of state intervention
(Ostrom, 2000; Gueth, 1995; Axelrod, 1986), most prominently on norms as a
result of repeated interactions between rational egoists and conditional cooperators
(Gueth, 1995; Axelrod, 1986) or between individuals that experience envy
and altruism (Teraji, 2007; Fehr and Gächter, 2001). However, we have very
limited knowledge whether individuals internalize sanctioned legal norms in the
context of political participation.

Many previous studies show that turnout tends to be higher in countries that
practice compulsory voting (Jackman, 1987; Blais and Young, 1996; Panagopoulos,
2011b). Most of these studies use national turnout data, which has-from a
methodological point of view-two major drawbacks. First, countries are different
in both their observed and unobserved factors. Second, countries are likely
to be confronted with different country-specific shocks over time. The chapter
advances existing approaches to compulsory voting in two dimensions. On the
one hand, data on Swiss cantonal is used to make a comparison. These units are
much more similar with regard to their covariates such as the institutional setting
and political culture. On the other hand, the analysis is based on a synthetic
control method (Abadie, Diamond, and Hainmueller, 2010), which is a recently
developed empirical strategy to chose a control group for a single-intervention
case in a data driven manner.

The results are important for the literature in political economy as participation
in referendums and elections constitutes "collective action par excellence"
(Shepsle and Bonchek, 1997, p. 251), as it is a very important means of citizens
in order to hold elected offiials accountable (Frey, 1994; Dal Bó, Foster, and
Putterman, 2010; Snyder and Strömberg, 2010; Lijphart, 1997). Furthermore,
from a normative point of view, some have argued that high levels of turnout
are desirable because they increase the legitimacy of the political system (Dahl,
1989; Lijphart, 1997; Franklin, 2004).

Chapter 3 tackles the question of how voters react to cost shocks in large
elections. I use data on individual turnout for over 80 referendum days and
combine it with data on the variation in the costs of voting, namely rainfall and
the introduction of postal voting. Rainfall exhibits considerable variation not
only between cantons but also between referendum days. The analysis pays
special attention to where individuals lie on the cost distribution. I make use
of detailed information on their average participation per 10 referenda which is
used as a measure of individual costs. Rainfall data has mainly been used as
an exogenous increase in the voting costs for aggregate data of US elections
(Knack, 1994; DiNardo, 2007). Another source of variation comes from the
sovereignty of Swiss cantons in terms of electoral law, both on the cantonal
and federal level. This autonomy created a situation where the earliest canton
(Basel-Town) introduced unconditional postal voting in 1978, whereas the last
canton (Ticino) did so in 2005. This variation has been used in Luechinger,
Rosinger, and Stutzer (2007) and Hodler, Luechinger, and Stutzer (2012) who
find considerable effects on aggregate turnout levels. Funk (2010) finds that the
effect of postal voting varies between smaller and bigger municipalities because
it not only lowers voting costs, but also removes social pressure that is higher in
small municipalities.

An important aspect of these changes in the costs to voting is the question
whether it leads to a change in the constituency that favors certain groups more
than others. In his presidential address, Lijphart (1997) expresses the concern
that low levels of turnout is a "serious democratic problem" since these translate
into a class-bias of political influence. The main argument put forward is that
disadvantaged people are less likely to vote which is aggravated when costs
to voting, such as diffcult registration procedures, are high (Wolfinger and
Rosenstone, 1980). Hansford and Gomez (2010) use the variation in rainfall as
an instrument for turnout and show that higher participation helps the Democrats.
While chapter 2 and 3 are related to cost shocks, chapter 4 focuses on
variation in mobilization power. An important component of democracy in
Switzerland is the fact that various proposals are at stake on the same referendum
day. As a consequence, most voters cast their vote for all proposals once they
have decided to turn out. This creates a situation in which the endogenous mobilization
mechanism is not working and relatively distant voters are mobilized
for proposals they would not vote on otherwise. Another effect is that highly
mobilizing proposals reduce the information level among voters. Therefore, it
is likely that this also affects the outcome of a vote. If it does, the timing of
referendums could be strategically influenced by the government based similar
to the timing of elections (Smith, 2003; Keppo, Smith, and Davydov, 2008).
More specifically, the existing literature predicts two sorts of effects coming
from additional relatively uninformed voters. First, the research on risk attitudes
has revealed that people with lower education and knowledge are more risk
averse (Frederick, 2005; Dohmen, Falk, Huffman, and Sunde, 2010; Dohmen,
Falk, Huffman, Sunde, Schupp, and Wagner, 2011), which would lead to higher
aggregate preferences for the status quo. Second, the literature on the European
Union has put forward the idea that referendums and elections are "secondorder"
national elections that are used as a means to punish the government (Reif
and Schmitt, 1980; Van der Eijk and Franklin, 1996; Garry, Marsh, and Sinnot,
2005). Both theoretical predictions will be examined empirically.