ArticlePDF Available

Abstract and Figures

In the case of coalition governments evidence suggests that voters have heuristics for assigning responsibility for economic outcomes to individual parties. Recent experimental evidence finds that the decision maker in a collective decision making entity with proposal power attracts a disproportionate amount of the blame or reward by those materially affected by these decisions. This essay demonstrates that voters identify the Prime Minister and the Finance Minister party as agenda setters on economic issues depending on whether the coalition context exaggerates or mutes the perceived agenda power of these parties. We define cabinet context as the extent to which coalition parties take issue ownership for particular policy areas. We find that when decision making is compartmentalized, voters perceive the finance minister as having agenda power and hence it receives a relatively larger economic vote; in more collective-like cabinet contexts the PM Party receives a larger economic vote.
Content may be subject to copyright.
Coalition Voting and the Economic Agenda Setter
Raymond M. Duch
Nuffield College
University of Oxford
raymond.duch@nuffield.ox.ac.uk
Albert Falc´o-Gimeno
Department of Political Science
University of Barcelona
afalcogimeno@ub.edu
October 2015
Abstract
In the case of coalition governments evidence suggests that voters have heuristics
for assigning responsibility for economic outcomes to individual parties. Recent exper-
imental evidence finds that the decision maker in a collective decision making entity
with proposal power attracts a disproportionate amount of the blame or reward by
those materially affected by these decisions. This essay demonstrates that voters iden-
tify the Prime Minister and the Finance Minister party as agenda setters on economic
issues depending on whether the coalition context exaggerates or mutes the perceived
agenda power of these parties. We define cabinet context as the extent to which coali-
tion parties take issue ownership for particular policy areas. We find that when decision
making is compartmentalized, voters perceive the finance minister as having agenda
power and hence it receives a relatively larger economic vote; in more collective-like
cabinet contexts the PM Party receives a larger economic vote.
1
1 Introduction
Government decisions in most democratic countries result from a process in which multiple
parties, that constitute the governing coalition, arrive at a collective decision. If voters
incorporate government performance into their vote preferences for particular parties, they
need to assign administrative responsibility to the individual parties making up the coalition.
Yet, very few models that purport to explain vote choice explicitly grapple with how voters
incorporate this multiparty feature of coalition governance into their vote calculus. In fact,
to the extent that the coalition context is given any consideration, it has been treated as
introducing “complexity” to the vote calculus that simply undermines the ability of voters
to hold incumbents accountable (Downs, 1957). By contrast, we argue that voters are
informed about the basic features of governing coalitions and they have heuristics for mapping
this basic information into more complex attribution of responsibility to governing parties.
Identifying these heuristics is the primary goal of this essay and doing so (or demonstrating
that no such heuristic exists) is an essential first step to understanding the nature of coalition-
directed voting.
It is true that coalition governments are, of course, quite diverse –they vary in terms of the
number of parties, their relative size, the nature of decision making processes within cabinet
deliberations, etc. And while the implications of this diversity has been explored in, for
example, models of coalition formation (Martin and Stevenson, 2001), duration, and policy
outputs (the size of government and of the government deficit, for example), the impact of
variations in these coalition characteristics on vote choice has been understudied. This essay
proposes to make a contribution to our understanding of how voters attribute responsibility
for government outcomes when they are the result of a collective decision taken by multiple
parties within a governing coalition. And in particular we are interested in identifying those
coalition decision making characteristics that affect how voters attribute responsibility to
the individual parties in the coalition.
Confronted with the collective decisions of a coalition government, rational voters are
2
expected to have a strategy for attributing responsibility to individual governing parties.
It is most likely the case that voters employ decision making short-cuts, or heuristics, to
assign responsibility (Duch, Przepiorka and Stevenson, 2014). Recent experimental evidence
suggests that agenda setting power represents one of the heuristics voters are likely to employ
for attributing responsibility for collective decisions (Duch, Przepiorka and Stevenson, 2014).
One goal of this essay is to better understand who the voter perceives as exercising agenda
power. Our empirical efforts here focus on voter responsibility attribution for policies that
shape the country’s economic performance. A clear candidate is the Prime Ministerial party
although one might also expect the party controlling the finance portfolio to be identified
as the proposer for policies that affect national economic performance. There is in fact
preliminary, comparative, evidence suggesting both the PM and Finance portfolio parties
have a disproportionately high economic vote (Duch and Stevenson, 2008).
But the magnitude of the Finance Minister Party’s economic vote (relative to the PM
Party) clearly varies. A recent study of the 2009 German election, for example, suggests the
economic vote was focused on the Chancellor’s party (CDU/CSU) with little evidence of an
economic vote for the Finance Minister Party (SPD) (Debus, Stegmaier and Tosun, 2014).
Our intuition here is that the coalition context shapes the public’s perception of proposal
power. As a result, the affect of the proposal power heuristic on economic voting will vary
by context.
We argue that the effect of the proposal power heuristic on economic voting will be
shaped by the extent to which the preferences of coalition parties are compartmentalized
(Falc´o-Gimeno, 2011, 2014). In highly ‘jurisdictional’ coalitions (i.e. partners with tangential
preferences) voters are more likely to perceive a particular party as the one to blame or give
credit to for the state of the economy, whereas in more ‘collective’ contexts (i.e. partners
with overlapped preferences) voters will recognize that responsibility is diffused which will
reduce the economic vote of coalition parties and particularly those with proposal power.
We provide observational evidence for 21 OECD countries from 1987 to 2010 testing this
3
argument.
The remainder of the paper is structured as follows. The next section briefly reviews the
literature on the attribution of responsibility for collective decisions. Section 3 presents our
argument and the hypotheses to be tested. In sections 4 and 5, we then describe our data,
discuss the methodology, and present results. Finally, the last section summarizes the main
contributions of the paper and concludes.
2 Responsibility attribution for collective decision mak-
ers: The role of heuristics
We are interested in better understanding how voters solve the responsibility attribution
challenge they confront when the incumbent choice is a government comprised of multiple
parties. The classic approach to this phenomenon, in the case of the economic vote, equates
multiple parties essentially with less, or worse, information about responsibility. As a result,
the economic vote is predicted to be more important when the institutional context concen-
trates responsibility for economic policy making on one or a few identifiable parties – the
so-called “high clarity” countries. In “low clarity” countries where responsibility is diffused
amongst multiple parties, the economic vote is expected to be attenuated. While empirical
support for these claims is considerable (for a review see Maravall, 2010), it provides little
insight into the micro-foundation of the coalition vote decision.
More recently there have been efforts to better understand the actual voting calculus
of voters in multi-party coalition contexts. Rather than simply assuming responsibility
attribution is attenuated when there are collective decisions (the “low clarity” cases), Duch
and Stevenson (2008) argue for a theoretical specification in which rational voters assign
varying degrees of responsibility to the different parties in a governing coalition. Duch
and Stevenson (2008, 253) conclude that many empirical results in the literature “leave
little room to doubt that the distribution of [policy making] responsibility is an important
4
factor conditioning both the magnitude of the economic vote across elections as well as its
distribution across parties”. They argue that economic voting is more important to the
electoral fortunes of individual parties on whom policy making responsibility is concentrated
than it is to those parties (in the same election) with less responsibility (Anderson, 1995;
Duch and Stevenson, 2008).1
Voting behavior scholars have recently provided a richer characterization of the voting
calculus in multi-party coalition government contexts. In particular, empirical findings sug-
gest that significant numbers of voters in coalitional contexts engage in “coalition directed
voting”, i.e. tactical voting for particular parties in order to try to bring a preferred coalition
to power. Kedar (2005) or Bargsted and Kedar (2009), for example, find that voters in con-
texts with coalition governments engage in compensational voting, i.e. certain voters vote
for more extreme parties with the goal of shifting the policy position of governing coalitions
closer to their ideal points. Based on data from 86 election surveys conducted in 22 countries,
Duch, May and Armstrong (2010) find that in 75% of these surveys more than 50% of voters
make coalition-directed calculations. However, in order to exercise these coalition-directed
votes, voters in coalitional contexts are faced with the challenge of mapping the observed
distribution of responsibility (namely, seats won and cabinet positions held) into actual
administrative responsibility within the cabinet. This presumes, first, that voters acquire
information about the composition of a coalition government and, second, they translate
observed characteristics of the governing coalition into shares of actual responsibility.
Determining how voters attribute shares of responsibility to the individual members of
coalition governments is an empirical challenge that has significant theoretical implications.
There have been some empirical insights. Anderson (2000) analyses aggregate election statis-
tics and finds that voters assign responsibility to the largest parties within the coalition: the
larger the party in terms of seats the more voters can assume it was responsible for policy
making. Duch and Stevenson (2008), on the other hand, find evidence for their proposition
1A similar, although differently motivated, argument is made by Anderson (1995).
5
that the proportion of cabinet portfolios signals the responsibility each party holds within a
coalition.
Recent experimental work on responsibility for collective decision making provides fur-
ther insights into voting for parties in coalition government contexts. Duch, Przepiorka and
Stevenson (2014) conduct lab experiments designed to isolate the heuristics individuals em-
ploy for holding decision makers accountable when decisions are made collectively (in their
case employing majority voting rules). Rather surprisingly, they find that neither voting
weights nor veto power is the heuristic individuals employ to punish unfair decisions. Their
somewhat unexpected result is that individuals disproportionately focus responsibility attri-
bution on the agenda setter – the individual with proposal power. The implication being
that voters will attribute responsibility to the party with proposal power in the governing
coalition. While surprising, the result accords well with evidence that agenda-setting power
influences outcomes in voting bodies (e.g. Weingast and Marshall, 1988; Cox and Magar,
1999) as well as policy outcomes in coalition governments (e.g. Laver and Shepsle, 1996).
In a coalition government, though, proposal power is not necessarily obvious. Who
manages the agenda? Who makes the proposals that the other partners are required to
accept, reject, or amend? Again, voters likely rely on heuristics. There is some evidence
suggesting that the Prime Ministerial (PM) party is perceived to have a disproportionate
influence on the decisions made within a coalition (Duch and Stevenson, 2008). But perhaps
the finance minister (FM) party should also be identified as the proposal maker on issues
concerning the economy (and hence the party to be rewarded or punished for the state of the
economy). Employing surveys from multiple countries and over an extensive time period we
identify the agenda setting heuristic employed by voters. Also, we focus on the fact that the
coalition decision making context differs considerably from one case to the next. Building
on Falc´o-Gimeno (2011, 2014) we describe how this cabinet context varies in a systematic
fashion and make the argument that variations in the coalition decision making context will
affect how voters employ the agenda setting heuristic when they exercise an economic vote.
6
3 Coalition governance and the economic vote: Hy-
potheses
We argue that voters have sufficient information to be able to identify the parties that are
economic policy agenda setters in a multiparty governing coalition. There is arguably a
distinct economic responsibility that is different from the overall level of responsibility that
each coalition partner has in the government. One obvious candidate for this economic
agenda setting role is the FM party. The expectation is that responsibility for coalition
decisions on economic issues would be attributed to the FM party. Our conjecture is that this
party is perceived as having agenda power on economic issues given that it routinely drafts
economic legislation and is typically the public spokesman for issues related to variations
in economic growth, trade balances, and budget balances. Voters are likely to learn about
the party’s economic decision making role because media representations of economic policy
making are likely to focus on the FM party. Accordingly, Duch, Przepiorka and Stevenson’s
(2014) agenda power experimental results suggest that voters are going to blame this party
in harsh economic times and reward it in times of prosperity.
But of course economic policy making is a central concern of the coalition government
as a whole and typically the PM party plays an active role both in economic policy mak-
ing and publicly defending, both nationally and internationally, the government’s economic
performance. Moreover, the PM in most cases controls the agenda of cabinet deliberations
and has strong influence over the legislative agenda. Our conjecture is that the PMs role in
coalition governance signals to the voters of the PM party’s agenda setting powers. In fact,
Duch and Stevenson (2013) provide evidence that the PM Party is perceived as the general
agenda setter in the minds of voters.
The PM and FM parties are likely to share agenda setting power over economic policy
making in the minds of the average voter. One of our goals is simply to explore the extent to
which voters use the agenda setting heuristic to attribute responsibility to the PM and FM
7
parties for economic outcomes. And to the extent possible, given the survey data available,
we attempt to tease out how much agenda setting power voters accord to the PM and FM
parties. Accordingly, we test the following hypothesis:
Hypothesis 1 All else equal, the state of the economy will affect the electoral fortunes of
the Prime Minister and Finance Minister parties within the coalition more
than the rest of the coalition partners.
A second key contribution of this essay is to demonstrate how the coalition context shapes
perceived agenda power. We conjecture that the extent to which parties can set the agenda
on economic issues depends on whether cabinets operate in a compartmentalized or collective
fashion. In a compartmentalized context, coalition parties decide policy in the jurisdictions
over which they have ministerial control without interfering in their partners’ domains. In a
collective context, policies are decided collectively by all partners in all dimensions, regardless
of the distribution of portfolios. Scholars disagree as to which of these styles prevail: Some
maintain that ministerial discretion is the rule and others consider it is the exception (see
for instance the debate between Warwick (1999b,a) and Laver and Shepsle (1999b,a)). Our
conjecture is that not all coalitions decide either one way or the other but, rather, some are
more prone to certain decision making arrangements and others to other forms of internal
governance.
Falc´o-Gimeno (2011, 2014) shows that for coalitions with members that have tangential
preferences a log-rolling of parties’ ideal points is a simple way to resolve partners’ differences
in emphasis (De Winter, 2002). That is, when each coalition member intensely cares about
a particular set of issues that do not overlap with each other, then compartmentalization
is preferable and more likely. Specialization of preferences therefore leads to the compart-
mentalization of decisions. By contrast, in cases where there is greater preference overlap
between partners, a collective decision making arrangement in all policy dimensions will lead
to results that partners prefer over compartmentalization.2
2It is beyond the scope of this paper to provide further explanation of this relationship between partners’
8
In sum, responsibility over economic issues in coalitions is not simply a matter of which
party occupies the Prime Ministership or controls the Finance ministry. Responsibility is also
conditioned by the decision making process. We argue that voters will more likely perceive
the PM or FM parties as agenda setters when the overall specialization/compartmentalization
of the coalition is high. By contrast, in coalitions where decisions are likely to be made
collectively, control over particular ministries will send a low quality signal about who is
responsible for the state of the economy. Voters will therefore be less inclined to consider
portfolio allocation a useful heuristic for responsibility attribution under these circumstances.
In a compartmentalized coalition, though, voters will have a crisper signal of agenda setting
powers for economic policy and will single out the PM and/or FM parties for responsibility
attribution.
Our second principal empirical undertaking will then test whether there is an interaction
between coalition context and the economic vote of the parties responsible for economic
policy agenda setting. Accordingly we test the following hypothesis:
Hypothesis 2 All else equal, the economic vote of the Prime and Finance minister parties
will be larger as the overall compartmentalization of the coalition increases.
4 Data and variables
The analysis builds on a number of data sets assembled by the authors. The period of
the analyses goes from 1987 to 2010. The dependent variable is the economic vote for spe-
cific parties in governing coalitions, our primary independent variables refer to the coalition
parties’ agenda setting powers on economic matters, and the main moderating regressor
measures the extent to which coalition members’ preference profiles are compartmentalized
or not.
preferences and coalition governance forms. Details are available in Falc´o-Gimeno (2011, 2014).
9
Dependent Variable: The Party Economic Vote. The Party Economic Vote (PEV) is
based on Duch and Stevenson (2008). They generate a measure of economic voting for each
governing coalition party that reflects the effect of perceptions of economic performance on
vote preference for these political parties. The estimates of PEV in this study updates their
original estimates using 297 voter preference surveys conducted in eighteen western democ-
racies from 1979-2010. The estimates are based on carefully specified statistical models of
individual voting behavior. Once obtaining estimates of the coefficients of well-specified vote
choice models, we use the estimated coefficients (and variance-covariance matrix) from the
model to produce predicted changes in support for each survey respondent when economic
perceptions became more negative by one unit. We also generate measures of uncertainty
around these predicted changes. The predicted changes in support for each party are av-
eraged over the sample to get an estimate of the average party economic vote (PEV). The
1,577 estimates obtained from these models (one for each party in the 297 voter preference
surveys) are our measure of the PEV in each voter preference survey. The values of the
variable should be read in the following terms: the more negative the values, the larger the
economic vote for a particular party. Of particular interest in our analysis will be the PEV
for the party holding the PM and Finance portfolio.3
Main Independent Variables: PM and FM. As our hypotheses suggest, the principal
independent variables should capture the extent to which each party in the governing coali-
tion is perceived as having proposal power over economic issues. Accordingly our measure
is simply whether or not the coalition party holds either the PM portfolio and the Finance
portfolio through two binary variables: PM Party and FM Party, respectively. The original
data comes from Seki and Williams’s (2014) Detailed Minister Summary dataset for the
period 1991-2012. This contains information on ministers whose tenure started as early as
1987 and links ministers to portfolios which allows us to identify the parties holding the PM
3The Rcode and original data for estimating these PEV is available from the authors. A description for
the estimation is available in Duch and Stevenson (2008).
10
and Finance ministries.
Moderating Variable: Compartmentalization. At the cabinet level, we measure the
extent to which government partners focus their attention and efforts on the same issues (i.e.
overlapped) as opposed to being “specialized” in their interests (i.e. compartmentalized).
This variable is a general measure of the extent to which policy preference profiles of the
parties sharing office are distinct: If the members of the coalition are intense in the same
policy areas, then their preferences will be overlapping. On the other hand, when the primary
policy concerns of coalition members do not coincide then we have a compartmentalized
cabinet. We say this is a moderating contextual variable because our conjecture is that the
effect of economic agenda power –i.e. the main independent variable– on the economic vote
–i.e. the dependent variable– will be affected by the extent to which coalition parties are in
general “policy distinct”.4We have argued that, on average, responsibility attribution for
(any) policy performance will be higher for a compartmentalized cabinet (see Falc´o-Gimeno,
2011, 2014).
Given that what we want to capture here is very much related to the salience approach
to policy preferences, this variable is measured as follows: We first take each party’s score
of salience for each preference category –namely, the share of the party election manifesto
that refers to each issue– from the version 2014b of the Manifesto Project dataset (Volkens
et al., 2014). We then group these very narrow CMP issue categories into the 13-policy
categories scheme built by B¨ack, Debus and Dumont (2011).5Since we need a government-
specific measure, we compute the cabinet standard deviation of these saliences for each
jurisdiction and finally average these standard deviations across the 13 jurisdictions. This
strategy produces an overall measure of the degree to which preferences are tangential or
distinctive in the coalition cabinet. The salience approach assumes that a higher average
4This is in fact very much related to Narud’s (1996) contention that different partners should be held
accountable for different policies depending on the extent to which voters relate government policy to the
programmatic commitment of certain parties.
5These categories are: Foreign affairs, interior, justice, finance, economy, defense, labor, education,
health, agriculture, industry, environment, and social affairs.
11
standard deviation represents a more compartmentalized coalition. This measure has already
been applied in the study of oversight mechanisms in coalition governments: Coalitions with
a large preference overlap (i.e. low compartmentalization) have been found to make more
extensive use of cross-partisan junior ministers (Falc´o-Gimeno, 2014).
To give a stylized example, suppose that only two policies exist: X and Y. In a coalition
cabinet formed by parties A and B, party A devotes 100% of its election programme to
policy X (and 0% to policy Y) while the stated preferences of party B are only related to
policy Y (0% to policy X and 100% to policy Y). The average policy standard deviation
–i.e. the compartmentalization measure– will be 70.7.6In another cabinet formed by parties
C and D that both care only about policy X (100% to policy X and 0% to policy Y), the
coalition would score 0 in the compartmentalization measure. In a third cabinet, party E
preference profile is the same as party A and party C, but party F cares equally about
both policies (50%-50% for X and Y). In such a case, the cabinet would score 35.4 in the
compartmentalization scale, lower than the A-B cabinet but higher than the C-D coalition.
Control Variables. As control variables we consider two other measures related to the
potential agenda setting capacity of the party on economic issues. First, the variable Party
Size (Gamson) refers to the relative seat share of each coalition party, i.e., its seat share
contribution to the coalition (these data are from D¨oring and Manow’s (2012) ParlGov
database).7Given the empirical strength of the Gamsons’ Law of proportionality between
seat shares and portfolio shares, this variable can also be interpreted as a proxy for the
proportion of ministries controlled by each party in the coalition. On the other hand, we
also explore the extent to which each parties’ economic profile in their public pronouncements
and campaign rhetoric affects their economic vote. The variable Economic Profile measures
the salience of the economy for each party in terms of the share of their manifesto referring
to economic issues –these data are from the Manifesto Project (Volkens et al., 2014). We
6Clearly, the score would also be 70.7 for any combination in which both parties coincide in both policies.
7We also employ this database to code our governments as coalitions or single-party cabinets.
12
Table 1: Descriptive Statistics
N Mean/Prop. S.D. Med.
Full Sample
Party Economic Vote 958 -0.001 0.031 0.001
Party in Government 912 0.356 0.479 0
Coalition 958 0.602 0.490 1
Coalition Parties
Party Economic Vote 252 -0.011 0.026 -0.005
NO PM & NO FM 252 0.508 0.501 1
NO PM & FM 252 0.099 0.300 0
PM & NO FM 252 0.159 0.366 0
PM & FM 252 0.234 0.424 0
Compartmentalization 226 3.022 1.370 3.165
Party Size (Gamson) 252 0.356 0.240 0.286
Economic Profile 228 0.093 0.049 0.085
follow B¨ack, Debus and Dumont’s (2011) strategy in measuring how much each coalition
party is interested in issues related to economic policies.
Table 1 provides summary information of our full sample, which contains a total of
958 party-government observations.8The relative proportion of government and opposition
parties is around 1
/32
/3, respectively. About 60% of our cases belong to contexts where a
coalition government is in office. When we restrict our sample to the coalition parties in
government, we have information for 252 parties –although we lose about 25 parties when
we incorporate the data from the Manifesto Project (Compartmentalization and Economic
Profile variables).
Regarding the dependent variable, we can see that, on average, the economic vote received
by a party in the full sample is practically zero. This is not surprising given that there are
both incumbent and opposition parties in the sample. However, if attention is focused on
8Our full sample includes governments from Australia, Austria, Belgium, Canada, Czech Republic, Esto-
nia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia, Lithuania, Luxem-
bourg, Netherlands, New Zealand, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, and the
United Kingdom. When we restrict our sample to coalition cases the number of countries obviously drops
to the 21 that a least had one coalition cabinet between 1987 and 2010.
13
coalition incumbent parties, the average economic vote goes down by a factor of 10 and the
standard deviation decreases.
With respect to the independent variables, we can see that the Prime Minister and the
Finance minister are frequently from the same party. Notwithstanding this general pattern,
16% of the coalition parties in our sample occupy the PM without controlling the Finance
Ministry. On the other hand, around 10% of our coalition parties control the Finance
Ministry but not the PM. In other words that means that a third of our FM parties did not
simultaneously occupy the Prime Ministership. That makes it possible to single out the FM
and PM agenda power heuristics and empirically evaluate the hypothesized FM effect on the
economic vote.
Regarding the main moderating variable, Table 1 shows that the average party in our
sample belongs to a cabinet that is more collective-like than compartmentalized (value 3 in
a range from 0 to 8.5), although there is substantial variation. Finally, our coalition parties
contribute to the cabinet with an average of around 36% of the total parliamentary seats of
the cabinet parties and devote around 9% of their election programmes to economic issues.
The statistical analyses shown in the next section come from the estimation of random
effects OLS linear regression models. To correct for the dependency between the units
from the same survey (i.e. the probabilities of voting for each party are presumably not
independent from each other), standard errors are clustered at the survey level. Later, we
also conduct a series of robustness checks where we test to what extent our findings are
robust to the control of country and time effects.
5 Results
Our point of departure in presenting the empirical results is to establish that the electoral
fortunes of government parties are in fact associated with the state of the economy, i.e.,
that there is in fact an economic vote in our sample. Table 2 presents four linear regression
14
Table 2: Strength of the Economic Vote and Type of Government
All Single-Party Coalition All
(1) (2) (3) (4)
Party In Government -0.026 -0.046 -0.019 -0.046
[0.003] [0.007] [0.002] [0.007]
Coalition Government -0.001
[0.002]
Party In * Coalition 0.027
[0.007]
Constant 0.008 0.009 0.008 0.009
[0.001] [0.002] [0.001] [0.001]
N 912 366 546 912
R20.160 0.238 0.153 0.206
Robust standard errors (clustered by survey) in brackets
analyses where the covariates are just the government/opposition status of the party and
the type of government (coalition vs single-party). Recall that the dependent variable here
is the magnitude of the economic vote for the individual parties in the coalition – the co-
efficient reflects the change in the vote probability for the party associated with a one unit
decline in subjective economic evaluations. Hence a negative coefficient for a given party
characteristic in the model indicates a larger PEV. Not surprisingly, the first two models re-
veal that incumbent parties receive a much stronger economic vote than opposition parties.
Models 2 and 3 run the same regression but in two restricted subsamples: cases under the
rule of a single-party government and coalition government cases. Here the evidence is again
consistent with what previous literature has found. The incumbent parties in single-party
governments have a significantly larger economic vote than those that share office in a multi-
party cabinet. The last model offers further confirmatory evidence that these differences are
actually statistically significant (as indicated by the coefficient estimate of the interaction).
Additionally, the models presented in this table indicate that the voters are holding the
15
collective “coalition” decision makers responsible for economic outcomes. Responsibility im-
plies that government parties in a coalition are punished for poor economic performance
and that the opposition typically benefits from a declining economy. Though weaker than
for single-party governments, models 3 and 4 show that this is the case. The coefficient
-0.019 in and the difference [-0.046+0.027], which are highly significant, confirm our expec-
tation that the coalition parties are being held responsible for economic outcomes: As the
economy worsens, incumbent coalition parties are punished. The positive and statistically
significant intercept suggests that non-incumbent parties benefit from bad economies. As we
expect, voters moderate their responsibility attribution for economic outcomes when there
is a multiparty government in office but it clearly persists.
5.1 Random Effects Models
As part of our effort to “unpack” this average collective responsibility attribution, we hy-
pothesized that the economic vote in coalitions will be stronger for certain parties than for
others. Table 3 further explores the accountability mechanism employed by voters for eco-
nomic outcomes under a coalition government, i.e. the sample is restricted to the cases in
which there is a coalition government. Models 1 and 2 show that, controlling for party size
and the extent to which the discourse of the party is attached to the economy, the party that
controls the Prime Ministership receives the largest economic vote. The coefficient for the
PM party dummy indicates that this party is disproportionately punished for bad economic
outcomes. Therefore, the evidence suggests, voters are inclined to treat the PM Party as the
agenda setter on economic matters, which is consistent with our theoretical expectations.
It does not appear to be the case, though, that the Finance Minister party is singled
out as responsible for economic outcomes. On average, the FM party appears not to receive
a significantly larger economic vote than junior coalition parties that neither control the
PM nor the FM (the reference category). In line with our expectations, the coefficient is
negative, but it does not reach conventional levels of statistical significance. Similarly, the
16
Finance Minister party’s economic vote is not statistically significant when we estimate it
independently from the PM party. The main effect of the FM Party variable in the interactive
Model 2 suggests that FM parties that do not control the PM do not receive a larger economic
vote than the rest of its partners. Neither is it the case that controlling the FM in addition
to controlling the Prime ministership significantly increases the economic vote above what a
party receives as a result of occupying the Prime ministership. This is what the negative but
statistically insignificant coefficient of the interaction PM * FM indicates. Occupying the
Prime ministership appears to provide voters with a strong signal as to who sets the agenda
for economic policy decisions, independently of whether the party also controls the finance
portfolio.
The results in these first two models also suggest that agenda power may not be the sole
heuristic for attributing responsibility in coalition governments. The significant coefficients
of the Party Size (Gamson) variable indicate that the parties that contribute to the coali-
tion with more parliamentary seat share – and presumably have more portfolios assuming
Gamson’s Law empirical regularity – have a larger economic vote. This is at odds with the
Duch, Przepiorka and Stevenson (2014) experimental results which suggest that there is a
largest decision maker effect but they do not find responsibility attribution proportional to
voting weight, which is how one might characterize shares of portfolios.
An additional factor that plays a role here is economic issue ownership. It contributes to
the voters’ perception of agenda power over economic policies. Coalition parties that signal
their association with financial issues receive a larger economic vote. Talking (or having
talked) in their manifestos about the economy when it is doing poorly negatively affects
parties’ electoral prospects.
These results suggest that there are parties in a governing coalition that can expect to
avoid being held responsible for economic outcomes. If an incumbent party provides no
signals regarding its influence over economic policy, it receives no punishment for a poorly
performing economy: Specifically, a coalition party that is neither in charge of the PM
17
nor the FM; a party that has few seats (“zero size”); and a party whose manifesto is not
associated with economic issues. This is implied by the positive and statistically significant
intercepts in the first two models.
The results presented in Models 1 and 2 do not take into account the coalition decision
making context. Our earlier conjecture is that features of coalition government decision
making can affect responsibility attribution. It might be the case, for example, that the
FM party proposal power effect varies by coalition decision making context. Accordingly,
Models 3 and 4 include the Compartmentalization variable. On average, and contrary to
our expectations, the members of a compartmentalized coalition (where it is clear which
issues are going to be managed by which party) receive a smaller economic vote than the
members of more “diffuse” coalitions. This is the implication of the positive coefficient for
compartmentalization in Model 3 (although it is not statistically significant). However, our
conjecture goes further than this “on average” correlation.
Our claim though is that compartmentalization conditions the extent to which coalition
parties are held responsible for economic outcomes. Accordingly, Model 4 in Table 3 ex-
plores whether the economic vote of the PM party and the FM party are conditioned on
compartmentalization. To accomplish this we include interactions PM * Compartment.,FM
* Compartment., and PM * FM * Compartment.. This specification suggests that in fact
the FM party in certain contexts is seen as having proposal power over the economy: the
FM party’s economic vote rises (becomes more negative) as the compartmentalization of the
coalition increases.
The results reported in Model 4 in Table 3 include three-way interactions. Accordingly,
to facilitate interpretation, Figure 1 provides an illustration of estimated substantive effects.9
As we would expect, compartmentalization has no impact on parties that control both the
FM and PM portfolios – their economic vote is constant across all values of compartmental-
ization. Compartmentalization clearly matters though for coalition parties that only control
9The predicted values are from Model 4 in Table 3.
18
Table 3: Strength of the Economic Vote in Coalitions
(1) (2) (3) (4)
PM Party -0.014 -0.013 -0.013 -0.018
[0.006] [0.006] [0.006] [0.013]
FM Party -0.004 -0.002 -0.002 0.021
[0.005] [0.005] [0.005] [0.008]
PM * FM -0.003 -0.002 -0.014
[0.007] [0.007] [0.019]
Compartmentalization 0.002 0.002
[0.001] [0.001]
PM * Compartment. 0.002
[0.003]
FM * Compartment. -0.008
[0.004]
PM * FM * Compartment. 0.005
[0.006]
Party Size (Gamson) -0.021 -0.021 -0.018 -0.019
[0.011] [0.011] [0.011] [0.010]
Economic Profile -0.073 -0.074 -0.084 -0.092
[0.027] [0.028] [0.029] [0.030]
Constant 0.009 0.009 0.005 0.003
[0.003] [0.003] [0.005] [0.004]
N 228 228 220 220
R20.203 0.204 0.198 0.214
Robust standard errors (clustered by survey) in brackets
19
Figure 1: Effect of Compartmentalization on Economic Vote
by Role in Coalition (Predicted Values)
|| ||||| | | || || ||||| | ||| || | || | |||| | || ||||| || || | | ||||| || || | || ||
-.06 -.04 -.02 0 .02
Strength of the Economic Vote
0 1 2 3 4 5 6 7 8 9
Compartmentalization
No PM, No FM No PM, FM
PM, No FM PM, FM
one of these two portfolios. If a party controls the FM portfolio but does not hold the Prime
Ministership, their economic vote is essentially zero in a coalition in which agenda setting
power is shared across coalition parties (a “diffuse” coalition). But as the agenda setting
power in the cabinet becomes more compartmentalized, the economic vote for the FM party
increases. Conversely, parties that hold the Prime Ministership but not the FM portfolio
have a high economic vote when agenda power is diffusely shared by coalition partners. And
as compartmentalization rises, their economic vote falls to a point of being insignificant from
zero. In coalitions that do not have a “super-party” that controls both the PM and the FM,
compartmentalization “transfers” the perception of the agenda power from the PM to the
FM. That is, in “diffuse” coalitions voters blame the PM party for poor economic outcomes,
while in compartmentalized coalitions voters cease to see the PM party as responsible and
focus on the FM party instead.
20
The two extreme cases in Figure 1 illustrate the importance of agenda power heuristics.
We see the largest economic vote accorded parties that control both the Prime Ministership
and the Finance Ministry in a coalition government. Moreover, this is not affected by the
degree of compartmentalization of the coalition. Voters consider these parties responsible for
the economy irrespective of the type of coalition. At the other extreme are coalition parties
that neither control the PM nor the FM. For these parties the economic vote is effectively
zero – compartmentalization has a weak effect that is not statistically significant.10
This essay proposes a novel view of voter reasoning in contexts where parties share
responsibility for policy making. We focus on one policy concern that shapes vote choice –
specifically, the economic vote. Does this novel agenda setting argument provide much added
values? If we ignore the agenda setting context then we have the conventional assessment
of the economic vote magnitude that is summarized in the first graph of Figure 2. Here we
simply estimate the economic vote for parties in single-party versus coalition government
contexts. Clearly parties that govern alone have a very large economic vote compared to
those who share power in a coalition government.
But our argument is that the small coalition party economic vote in the first frame of Fig-
ure 2 masks considerable contextual variation in responsibility attribution. Our contention
is that voters employ information and decision-making short-cuts in order to determine how
to allocate responsibility amongst parties in a governing coalition. Once we incorporate the
agenda setting heuristic into the vote utility function, we obtain a much richer characteriza-
tion of the economic vote. The remaining frames of Figure 2 illustrate the extent to which the
economic vote is condition on the agenda setting heuristic. A party’s agenda setting power
is determined by the portfolios it controls and the cabinet’s decision making process.11
The second frame illustrates the importance of cabinet compartmentalization. When
10It is true that, for the sake of presentability, Figure 1 does not provide confidence intervals for the
predicted values and that the compartmentalization values for which the FM party heuristic is strongest are
not very common in our sample. This is why Figure 2 shows the predicted economic vote with confidence
intervals for various types of parties under “reasonable” values of the moderating variable.
1110th (low), 50th (medium), and 90th (high) percentiles of the variable compartmentalization correspond
to the values 1.45, 3.17, and 4.51, respectively.
21
policy decision making in a cabinet is diffuse such that no single party has ownership for
particular policy areas, responsibility attribution is focused on the PM party. And in our
analysis, the PM party has a large economic vote while the FM party’s economic vote is
small. In fact, the electoral punishment of the PM party in a coalition (and in particular if it
is simultaneously in charge of the finance ministry) is not that different from the incumbent
party in a single-party government.
The third and fourth frames illustrate how an increasing compartmentalization of cabinet
decision making increases the economic vote of parties controlling the Finance Ministry. In
the third frame where compartmentalization assumes moderate values, parties controlling the
Finance Minister, but not controlling the Prime Ministership, see a quite dramatic increase
in their economic vote. Finally in the fourth frame where compartmentalization is high,
controlling the FM or the PM appears to have similar implications for a party’s economic
vote. As a result, the economic vote for parties controlling the FM but not the PM is
high, and similar to the economic vote for parties controlling the PM but not the FM. In
fact, the economic vote for these two cases is similar to that of parties controlling both the
PM and FM portfolios. Clearly the agenda setting heuristic matters for vote choice and
it is conditioned, as expected, on the compartmentalization of the cabinet decision making
process.12
12In the Online Appendix I we also illustrate how an increasing compartmentalization of cabinet decision
making increases the economic vote of parties controlling the Finance Ministry.
22
Figure 2: Economic Vote by Type of Government and Role in Coalition
(Predicted Values and 95% CI)
S-P.~&~Opp.
S-P.~&~Gov.
Coal.~&~Opp.
Coal.~&~Gov.
General
-.05 -.04 -.03 -.02 -.01 0 .01
No PM~&~No FM
No PM~&~FM
PM~&~No FM
PM~&~FM
Low Compartment. (Coalition)
-.05 -.04 -.03 -.02 -.01 0 .01
No PM~&~No FM
No PM~&~FM
PM~&~No FM
PM~&~FM
Median Compartment. (Coalition)
-.05 -.04 -.03 -.02 -.01 0 .01
No PM~&~No FM
No PM~&~FM
PM~&~No FM
PM~&~FM
High Compartment. (Coalition)
-.05 -.04 -.03 -.02 -.01 0 .01
23
5.2 Robustness Checks
It could be argued that a substantial part of the variation in our dependent variable is
the result of unaccounted for country-level heteregeneity. Our assumption is that there is
considerable within-country variation in compartmentalization and it is this within-country
variation that provides leverage in identifying the compartmentalization effect. Similarly,
the influence of time trends could also undermine the robustness of the main findings. The
economy could be of increasing or decreasing importance over time in the vote utility function
and this trend could be confounded with trends in comparmentalization. This could be either
a general tendency or a country-specific one.
Table 4 presents the results of a series of robustness tests designed to address these issues.
Models 1 to 5 take the full random effects model and add country fixed effects, a common
time trend, and country-specific time trends in different combinations. A comparison of these
models to the last model in Table 3 indicates only small changes in the results. In particular,
note that the coefficient estimate for the interaction FM * Compartment. is always negative
and statistically significant at a 95% confidence level. This confirms that the higher cabinet
compartmentalization results in a larger economic vote for the FM party.
Table 5 indicates that our results are also robust to dropping one country at a time. In
particular, the effect of the variable Party FM conditioned on the values of Compartmen-
talization, i.e., FM * Compartment., is essentially the same regardless of which country we
exclude. Compared to coefficient estimation of the full baseline model with country fixed
effects and country-specific time trends (-0.012), the coefficients for the different exclusions
range from -0.016 (when Belgium is dropped) to -0.009 (when Italy is excluded). The mag-
nitude of the standard errors is very similar and they all reach statistical significance, with
the highest p-value being 0.09.
Our main findings are therefore robust to a variety of alternative model specifications;
they are not dependent on one country or the other driving the results; the principal effects
are not confounded with temporal trends; and the main effects are not confounded with
24
Table 4: Full models of Table 3 (Robustness to Country and Time Effects)
(1) (2) (3) (4) (5)
PM Party -0.013 -0.016 -0.011 -0.011 -0.006
[0.014] [0.012] [0.013] [0.013] [0.013]
FM Party 0.020 0.022 0.022 0.022 0.029
[0.010] [0.008] [0.010] [0.010] [0.012]
PM * FM -0.021 -0.017 -0.027 -0.027 -0.041
[0.020] [0.018] [0.019] [0.019] [0.023]
Compartmentalization 0.004 0.002 0.004 0.004 0.004
[0.001] [0.001] [0.001] [0.001] [0.002]
PM * Compartment. -0.001 0.001 -0.001 -0.001 -0.003
[0.004] [0.003] [0.003] [0.003] [0.003]
FM * Compartment. -0.009 -0.009 -0.010 -0.010 -0.012
[0.004] [0.004] [0.004] [0.004] [0.005]
PM * FM * Compartment. 0.010 0.005 0.011 0.011 0.015
[0.007] [0.006] [0.007] [0.007] [0.007]
Party Size (Gamson) -0.024 -0.018 -0.023 -0.023 -0.022
[0.011] [0.010] [0.011] [0.011] [0.012]
Economic Profile -0.070 -0.094 -0.076 -0.076 -0.074
[0.034] [0.030] [0.033] [0.033] [0.038]
Constant 0.002 0.696 1.204 1.200 -0.101
[0.009] [0.382] [0.494] [0.493] [1.447]
Country Fixed Effects yes no yes no yes
Common Time Trend no yes yes no no
Country-specific Time Trend no no no yes yes
N 220 220 220 220 220
R20.297 0.221 0.312 0.312 0.359
Robust standard errors (clustered by survey) in brackets
25
Table 5: Effect of FM * Compartment. (Robustness to Removal of One Country at a Time)
Removed Country coef. s.e. N R2
Baseline Model -0.012 0.005 220 0.359
(all countries)
Ireland -0.014 0.006 205 0.359
Netherlands -0.012 0.005 203 0.372
Belgium -0.016 0.006 188 0.395
Luxembourg -0.011 0.005 206 0.428
France -0.012 0.005 211 0.337
Germany -0.009 0.005 193 0.359
Poland -0.011 0.005 212 0.373
Austria -0.012 0.005 218 0.354
Hungary -0.010 0.005 203 0.320
Czech Rep. -0.012 0.005 205 0.324
Slovakia -0.011 0.005 200 0.362
Italy -0.009 0.005 202 0.364
Slovenia -0.012 0.005 218 0.358
Estonia -0.010 0.005 214 0.376
Lithuania -0.012 0.005 218 0.374
Finland -0.013 0.006 213 0.361
Iceland -0.012 0.005 218 0.357
Japan -0.012 0.005 219 0.353
Australia -0.011 0.005 214 0.361
Baseline model (all countries) with country fixed effects and
country-specific time trends.
26
unmeasured country-level heterogeneity.13
6 Conclusions
Duch and Stevenson (2008) specified a vote utility function in which responsibility attribu-
tion for economic outcomes is conditioned on a party’s competency for economic outcomes.
This implies that in the case of coalition governments voters have some heuristic for assigning
responsibility for economic outcomes to individual parties. Employing experimental meth-
ods, Duch, Przepiorka and Stevenson (2014) suggest that one important heuristic is agenda
power. The decision maker in a collective decision making entity with proposal power tends
to attract a disproportionate amount of the blame or reward by those who are materially
affected by these decisions. As they point out in their essay, it is unclear which decision
making party in a coalition will be perceived as having agenda power over economic policy
decisions. They provide some evidence that voters consider the PM Party to be the agenda
setter.
This essay provides an empirical foundation for the Duch, Przepiorka and Stevenson
(2014) conjecture regarding the agenda setting heuristic. It makes two contributions to
our understanding of the responsibility attribution heuristic voters employ for collective
decisions. First, we advance our understanding of the specific coalition parties voters treat
as having agenda setting power. The Duch, Przepiorka and Stevenson (2014) experimental
results suggest that a party’s ability to control the agenda for economic outcomes should
contribute to whether voters hold them responsible for economic outcomes. Our results
suggest that the PM party is certainly perceived as having agenda power and, under certain
13We have conducted a series of additional robustness checks in the Online Appendix I. We intro-
duce controls involving the interaction of our main moderating variable (compartmentalization) and the
size and squared size of each party. We are able to rule out that the significant interaction we observe
FM*Compartment. is because compartmentalized coalitions form in contexts where the members of the
coalition are very different in terms of size. We also introduce controls involving the coalition variable Party
Size Imbalance, which measures the degree of disparity between the party sizes of the members of the coali-
tion. We are able to rule out the likelihood that the significant interaction we observe FM*Compartment. is
because compartmentalized coalitions form in contexts where the members of the coalition are very different
in terms of size.
27
circumstances, the Finance party also fit this role.
Our second contribution is particularly novel in that it explores how the coalition context
can condition responsibility attribution. We argue that there are features of coalitions’
decision making that can either exaggerate or mute responsibility attribution accorded to
the agenda setters. Building on Falc´o-Gimeno (2011, 2014) we argue that cabinet decision
making can be structured in a compartmentalized fashion where coalition parties take issue
ownership for particular policy area s. In this case we conjecture that proposal power is
enhanced – it is more visibility associated with a particular party – and we expect to see
agenda setters receiving a relatively large economic vote. By contrast, in the case where
coalition decision making is collective and parties have an overlapping association with policy
areas, proposal power is reduced which results in a smaller economic vote.
Our empirical results confirm the initial conjecture but in a rather unexpected way. In
all coalitions, irrespective of the particular context, the Prime Ministerial party receives
the largest economic vote, especially when it simultaneously occupies the Finance ministry.
Cabinet contexts, in which responsibility is compartmentalized, provide voters with a signal
regarding the agenda power of the party controlling the Finance Ministry. Accordingly, we
find that in compartmentalized coalitions Finance Minister parties that do not control the
PM have a high economic vote. In fact, it appears that as compartmentalisation rises, voters’
perception of the Finance Minister party’s economic responsibility increases possibly at the
expense of the Prime Minister party’s responsibility for the economy.
28
References
Anderson, Christopher J. 1995. “The dynamics of public support for coalition governments.”
Comparative Political Studies 28(3):350–383.
Anderson, Christopher J. 2000. “Economic voting and political context: A comparative
perspective.” Electoral Studies 19(2-3):151–170.
ack, Hanna, Marc Debus and Patrick Dumont. 2011. “Who gets what in coalition govern-
ments? Predictors of portfolio allocation in parliamentary democracies.” European Journal
of Political Research 50(4):441–478.
Bargsted, Matias and Orit Kedar. 2009. “Coalition-Targeted Duvergerian Voting: How
Expectations Affect Voter Choice under Proportional Representation.” American Journal
of Political Science 53:307–23.
Cox, Gary W and Eric Magar. 1999. “How much is majority status in the US Congress
worth?” American Political Science Review 93:299–310.
De Winter, Lieven. 2002. Parties and Government Formation, Portfolio Allocation, and
Policy Definition. In Political parties in the new Europe: Political and analytical challenges,
ed. Kurt R. Luther and Ferdinand M¨uller-Rommel. New York, NY: Oxford University
Press.
Debus, Marc, Mary Stegmaier and Jale Tosun. 2014. “Economic Voting under Coalition
Governments: Evidence from Germany.” Political Science Research and Methods 2(1):49–
67.
oring, Holger and Philip Manow. 2012. “Parliament and government composition database
(ParlGov).” An infrastructure for empirical information on parties, elections and govern-
ments in modern democracies. Development Version.
29
Downs, Anthony. 1957. An Economic Theory of Political Action in a Democracy. Vol. 65
New York: Harper and Row.
Duch, Raymond M., Jeff May and David Armstrong. 2010. “Coalition-Directed Voting in
Multi-Party Democracies.” American Political Science Review 104(4):698–719.
Duch, Raymond M. and Randolph T. Stevenson. 2008. The economic vote: How political
and economic institutions condition election results. Cambridge: Cambridge University
Press.
Duch, Raymond M., Wojtek Przepiorka and Randolph T. Stevenson. 2014. “Responsibil-
ity Attribution for Collective Decision Makers.” American Journal of Political Science
(forthcoming).
Duch, Raymond and Randy Stevenson. 2013. “Voter Perceptions of Agenda Power and
Attribution of Responsibility for Economic Performance.” Electoral Studies (32):512–516.
Falc´o-Gimeno, Albert. 2011. Coalition Governance: Causes and Consequences PhD thesis
Universitat Pompeu Fabra.
Falc´o-Gimeno, Albert. 2014. “The use of control mechanisms in coalition governments: The
role of preference tangentiality and repeated interactions.” Party Politics 20(3):341–356.
Kedar, Orit. 2005. “When Moderate Voters Prefer Exterme Parties: Policy Balancing in
Parliamentary Elections.” American Political Science Review 99(2):185–199.
Laver, Michael J. and Kenneth A. Shepsle. 1996. Making and breaking governments: Cabinets
and legislatures in parliamentary democracies. Cambridge: Cambridge University Press.
Laver, Michael and Kenneth A Shepsle. 1999a. “Government formation and survival: A
rejoinder to Warwick’s reply.” British Journal of Political Science 29(2):412–415.
Laver, Michael and Kenneth A Shepsle. 1999b. “Understanding government survival: empir-
ical exploration or analytical models?” British Journal of Political Science 29(2):395–401.
30
Maravall, Jos´e M. 2010. “Accountability in coalition governments.” Annual Review of Po-
litical Science 13:81–100.
Martin, Lanny and Randy Stevenson. 2001. “Government Formation in Parliamentary
Democracies.” American Journal of Political Science 45:33–50.
Narud, Hanne Marthe. 1996. “Party Policies and Government Accountability.” Party Politics
2(4):479–506.
Seki, Katsunori and Laron K. Williams. 2014. “Updating the Party Government Data Set.”
Electoral Studies 34(2):270–279.
Volkens, Andrea, Onawa Lacewell, Pola Lehmann, Sven Regel, Henrike Schultze
and Annika Werner. 2014. The Manifesto Data Collection. Manifesto Project
(MRG/CMP/MARPOR). Version 2014b. Berlin: Wissenschaftszentrum Berlin f¨ur Sozial-
forschung (WZB).
Warwick, Paul V. 1999a. “Getting the assumptions right: a reply to Laver and Shepsle.”
British Journal of Political Science 29(2):402–412.
Warwick, Paul V. 1999b. “Ministerial autonomy or ministerial accommodation? Contested
bases of government survival in parliamentary democracies.” British Journal of Political
Science 29(2):369–394.
Weingast, Barry R. and William J. Marshall. 1988. “The industrial organization of Congress;
or, why legislatures, like firms, are not organized as markets.” The Journal of Political
Economy 96:132–163.
31
ONLINE APPENDIX I
The Online Appendix I provides additional views of the analyses provided in the main
text along with a set of additional robustness checks on the model specifications.
Figure A.1 reproduces Figure 2 in the main text. It shows the predicted economic vote
with confidence intervals for various types of parties under “reasonable” values of the moder-
ating variable. The frames in Figure A.1 illustrate how an increasing compartmentalization
of cabinet decision making increases the economic vote of parties controlling the Finance
Ministry.
i
Figure A.1: Original Figure 2. Predicted Values and 95% CI
S-P.~&~Opp.
S-P.~&~Gov.
Coal.~&~Opp.
Coal.~&~Gov.
General
-.05 -.04 -.03 -.02 -.01 0 .01
No PM~&~No FM
No PM~&~FM
PM~&~No FM
PM~&~FM
Low Compartment. (Coalition)
-.05 -.04 -.03 -.02 -.01 0 .01
No PM~&~No FM
No PM~&~FM
PM~&~No FM
PM~&~FM
Median Compartment. (Coalition)
-.05 -.04 -.03 -.02 -.01 0 .01
No PM~&~No FM
No PM~&~FM
PM~&~No FM
PM~&~FM
High Compartment. (Coalition)
-.05 -.04 -.03 -.02 -.01 0 .01
ii
Figure A.2 is an alternative presentation of Figure A.1. The panels now refer to the type
of coalition party and within each panel we have different degrees of compartmentalization.
Figure A.2: Alternative presentation of Figure 2 from main text. Predicted Values and 95% CI
S-P.~&~Opp.
S-P.~&~Gov.
Coal.~&~Opp.
Coal.~&~Gov.
All Parties
-.05 -.04 -.03 -.02 -.01 0 .01
Low Comp.
Med. Comp.
High Comp.
Coal. Party (NOPM&NOFM)
-.05 -.04 -.03 -.02 -.01 0 .01
Low Comp.
Med. Comp.
High Comp.
Coal. Party (NOPM&FM)
-.05 -.04 -.03 -.02 -.01 0 .01
Low Comp.
Med. Comp.
High Comp.
Coal. Party (PM&NOFM)
-.05 -.04 -.03 -.02 -.01 0 .01
Low Comp.
Med. Comp.
High Comp.
Coal. Party (PM&FM)
-.05 -.04 -.03 -.02 -.01 0 .01
iii
Table A.1 replicates Table 3 from the main text.
Table A.1: Table 3 from main text.
(DV: Economic Vote)
(1) (2) (3) (4)
PM Party -0.014 -0.013 -0.013 -0.018
[0.006] [0.006] [0.006] [0.013]
FM Party -0.004 -0.002 -0.002 0.021
[0.005] [0.005] [0.005] [0.008]
PM * FM -0.003 -0.002 -0.014
[0.007] [0.007] [0.019]
Compartmentalization 0.002 0.002
[0.001] [0.001]
PM * Compartment. 0.002
[0.003]
FM * Compartment. -0.008
[0.004]
PM * FM * Compartment. 0.005
[0.006]
Party Size (Gamson) -0.021 -0.021 -0.018 -0.019
[0.011] [0.011] [0.011] [0.010]
Economic Profile -0.073 -0.074 -0.084 -0.092
[0.027] [0.028] [0.029] [0.030]
Constant 0.009 0.009 0.005 0.003
[0.003] [0.003] [0.005] [0.004]
N 228 228 220 220
R20.203 0.204 0.198 0.214
Robust standard errors (clustered by survey) in brackets
iv
Table A.2 provides further robustness checks on the model results from Table A.1. The
first model in this table is the full model in Table A.1. The rest of the columns introduce
controls involving the interaction of our main moderating variable (compartmentalization)
and the size and squared size of each party. This is to rule out that the significant interaction
we observe FM*Compartment. is because compartmentalized coalitions form in contexts
where the members of the coalition are very different in terms of size. These results suggest
that controlling for Gamson and Gamson2does not affect the strength of the interaction
FM*Compartment..
v
Table A.2: Further checks for models in Table 3 of main text (I).
(DV: Economic Vote)
(1) (2) (3) (4) (5) (6)
PM Party -0.018 -0.013 -0.014 -0.023 -0.023 -0.024
[0.013] [0.006] [0.006] [0.013] [0.013] [0.013]
FM Party 0.021 -0.002 -0.003 0.019 0.019 0.018
[0.008] [0.005] [0.005] [0.008] [0.008] [0.008]
PM * FM -0.014 -0.003 0.002 -0.014 -0.014 -0.013
[0.019] [0.007] [0.007] [0.019] [0.018] [0.019]
Compartmentalization 0.002 0.003 0.003 0.004 0.003 0.004
[0.001] [0.002] [0.001] [0.002] [0.001] [0.004]
PM * Compartment. 0.002 0.004 0.003 0.004
[0.003] [0.004] [0.004] [0.004]
FM * Compartment. -0.008 -0.008 -0.008 -0.008
[0.004] [0.004] [0.004] [0.003]
PM * FM * Compartment. 0.005 0.005 0.006 0.006
[0.006] [0.007] [0.006] [0.007]
Party Size (Gamson) -0.019 -0.005 0.020 0.001 0.017 0.027
[0.010] [0.019] [0.026] [0.021] [0.026] [0.082]
Party Size * Compartment. -0.004 -0.008 -0.003
[0.006] [0.007] [0.026]
Party Size (Gamson)2-0.031 -0.018 -0.029
[0.040] [0.044] [0.112]
Party Size2* Compartment. -0.005 -0.010 -0.006
[0.008] [0.010] [0.036]
Economic Profile -0.092 -0.083 -0.083 -0.090 -0.090 -0.090
[0.030] [0.030] [0.030] [0.030] [0.030] [0.030]
Constant 0.003 -0.001 -0.005 -0.002 -0.005 -0.007
[0.004] [0.006] [0.006] [0.006] [0.006] [0.014]
Observations 220 220 220 220 220 220
R20.214 0.200 0.208 0.218 0.226 0.226
Robust standard errors (clustered by survey) in brackets
vi
Table A.3 provides further robustness checks on the model results from Table A.1. The
first model in this table is again the full model from Table A.1. The rest of columns in-
troduce controls involving the coalition variable Party Size Imbalance, which measures the
degree of disparity between the party sizes of the members of the coalition. This is again
to rule out the likelihood that the significant interaction we observe FM*Compartment. is
because compartmentalized coalitions form in contexts where the members of the coalition
are very different in terms of size. These results suggest that controlling for Party Size Im-
balance does not affect the strength of the interaction FM*Compartment.. It is neither the
case that Party Size Imbalance reflects the same context than Compartmentalization: the
interaction FM*PartySizeImbalance, though negative, does not reach statistical significance
as FM*Compartment. does.
vii
Table A.3: Further checks for models in Table 3 of main text (II).
(DV: Economic Vote)
(1) (2) (3)
PM Party -0.018 -0.016 -0.011
[0.013] [0.012] [0.009]
FM Party 0.021 0.017 0.006
[0.008] [0.008] [0.009]
PM Party * FM Party -0.014 -0.011 0.010
[0.019] [0.019] [0.012]
Compartmentalization 0.002 0.002 0.001
[0.001] [0.001] [0.001]
PM Party * Compartment. 0.002 0.000
[0.003] [0.003]
FM Party * Compartment. -0.008 -0.007
[0.004] [0.004]
PM Party * FM Party * Compartment. 0.005 0.006
[0.006] [0.006]
Party Size Imbalance -0.018 0.007
[0.010] [0.011]
PM Party * Party Size Imbalance -0.042
[0.043]
FM Party * Party Size Imbalance -0.062
[0.055]
PM Party * FM Party * Party Size Imbalance 0.016
[0.058]
Party Size (Gamson) -0.019 -0.021 -0.004
[0.010] [0.010] [0.009]
Economic Profile -0.092 -0.061 -0.066
[0.030] [0.030] [0.030]
Constant 0.003 0.006 -0.002
[0.004] [0.004] [0.006]
Observations 220 216 216
R20.214 0.225 0.249
Robust standard errors (clustered by survey) in brackets
viii
... Hence while it is true that coalition partners govern jointly and that the cabinet is characterized by intra-party bargaining and coordination ( Müller and Strøm, 2000;Thies, 2001;Kim and Loewenberg, 2005; Vanberg, 2011, 2014), the access to policy-specific expertise grants ministers considerable policy prerogatives. Given the central role of ministerial portfolios, if voters understand specific party responsibilities within the coalition, then when it comes to the performance vote, perhaps they will hone in on the party in control of specific ministries ( Duch and Falcó-Gimeno, 2015). If so, when it comes to specific issues, voters are expected to reward or punish not the party of the head of the cabinet or simply the party that controls more ministerial posts, but rather the party in charge of the ministries that deal with those specific issues ( Narud and Valen, 2008). ...
Article
How do voters attribute responsibility for government outcomes when they are the result of a collective decision taken by multiple parties within a coalition government? In this article we test the argument that in a multiparty coalition system, responsibility attribution should vary according to the quantity and quality of portfolios that the coalition partner controls. The article uses data from the Italian National Election Study in Italy, a country usually characterized by governments formed by more than two parties. We find no consistent empirical evidence that coalition parties collectively suffer from perceived negative performance. While the prime minister party is held responsible on average more than the other coalition partners, responsibility attribution decreases by party size in the parliament rather than by the quantity of ministerial portfolios the incumbent party controls. Issue saliency, however, plays an important role in the retrospective voting mechanism. These results have important implications for our understanding of electoral behaviour and democratic accountability.
Article
Full-text available
While there is plenty of empirical research on the determinants of voting behaviour in German elections, studies that focus on the perceived economic performance between two elections on voter’s decision making are scare. In this paper, we shed light on the impact of economic voting in federal elections for the German parliament. We argue that, first, economic voting should have a decisive impact on voting behaviour if the election campaign was dominated by economic issues and if the macro-economic situation became worse. Secondly, we combine theories of coalition politics and cabinet decision-making with theoretical approaches on voting behaviour and argue that voters punish in particular those government parties that were in charge of key cabinet offices related with economic issues. Thirdly, we expect that these effects remain significant after controlling for a whole battery of additional factors. The empirical results, which are based on data from national election studies in the time period between 1987 and 2009, support our key arguments.
Book
This book proposes a selection model for explaining cross-national variation in economic voting: Rational voters condition the economic vote on whether incumbents are responsible for economic outcomes, because this is the optimal way to identify and elect competent economic managers under conditions of uncertainty. This model explores how political and economic institutions alter the quality of the signal that the previous economy provides about the competence of candidates. The rational economic voter is also attentive to strategic cues regarding the responsibility of parties for economic outcomes and their electoral competitiveness. Theoretical propositions are derived, linking variation in economic and political institutions to variability in economic voting. The authors demonstrate that there is economic voting, and that it varies significantly across political contexts. The data consist of 165 election studies conducted in 19 different countries over a 20-year time period.
Article
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Article
Research on government formation in parliamentary democracies is replete with theoretical arguments about why some coalitions form while others do not. Unfortunately, this theoretical richness has not led to the development of an empirical tradition that allows scholars to evaluate the relative importance of competing theories. We resolve this problem by applying an empirical framework that is appropriate for modeling coalition choice to evaluate several leading explanations of government formation. Our approach allows us to make conclusions about the relative importance of traditional variables relating to size and ideology and to assess the impact of recent new-institutionalist theories on our ability to explain and predict government formation.
Article
In this article I argue that coalitions will tend to employ control mechanisms to facilitate the adoption of compromise policies only when the expected benefit of their use is high enough. When partners are already satisfied with log-rolling policies (compartmentalized by jurisdiction), or when compromise is already attainable self-enforcingly, there are few incentives to use them. Conversely, when partners are interested in compromise policies but are unable to reach that outcome in equilibrium, then control mechanisms are likely to be implemented. The empirical evidence offered tends to support the two main hypotheses of this work: control mechanisms are less necessary when the tangentiality of partners' preferences is high and when they foresee frequent mutual interactions. However, that seems to work better for the allocation of watchdog junior ministers rather than for the writing of comprehensive policy agreements.
Article
Scholars have identified the crucial role of government characteristics in studies of political behavior, comparative institutions, and political economy. An invaluable data source for government characteristics is the Woldendorp et al. (2000)Party Government in 48 Democracies data set. We describe our update to this data set from the late-1990s through 2011. We then present a variety of additions to the data set that are intended to increase its usage by reducing the obstacles associated with using the data in conjunction with other popular data sets. We illustrate the utility of this update by providing a variety of means of conceptualizing government stability.
Article
In two recent experiments (one in the lab and one over the internet) concerning collective decision making we determined that individuals mainly assign responsibility to the decision maker with agenda power and with the largest vote share. We found rather weak evidence that responsibility is assigned to decision makers with veto power or allocated proportional to weighted voting power. Our conjecture then is that individuals in our online experiment who recognized the importance of proposal power in the embedded experiment will be those more likely to exercise an economic vote for the Conservative PM Party (since they are the agenda setter in the governing coalition) and for the opposition Labour Party. The conjecture is confirmed. Essentially, the data show that economic voting at the individual level is confined to individuals who understand the value of proposal power. This in turn suggests that the economic vote itself is motivated by a coherent attempt to punish or reward parties that actually deserve it – in the specific sense that they were mostly responsible for choosing the policies that were implemented. Further, the strong reliance on proposal power as the workhorse of this mechanism of accountability, tells us that simple heuristics can do a lot of the work that cold rationality and complex calculation have done in much of the previous discussion of economic voting.
Article
We implemented two experiments that are designed to recover the responsibility attribution reasoning of individuals. The rst endeavor is a lab experiment designed to identify what decision maker attributes, of the actors in a collective decision making group, affect responsibility attribution. Here we simply have individuals identify the decision makers who they want to punish. The collective decision-makers individuals chosen to punish are not necessarily those decision makers individuals considered to have hadthe most influence on the decision making outcome. Our intuition is that this is the case but we designed a second, internet, experiment, with a more representative sample of the UK population, that helps answer this question. In this second experiment individuals are expected to anticipate the relative impact of dierent decision makers on collection
Article
For some time now, formal modelling has been touted by its supporters as a panacea for political science – or at least as a major step forward in the discipline's development. Certainly, it embodies a number of praiseworthy elements. Its insistence on starting with a parsimonious and precisely formulated set of assumptions cannot help but constrain slippery thinking, for example, and its rigorous working out of implications, while often demonstrating the obvious, occasionally leads to unanticipated and intriguing results. Moreover, the combination of precision and rigour holds forth the promise of generating relatively clear-cut tests of rival explanations, a major boon – if it proves true – in a discipline more inclined to abandon theories than to disconfirm them. How much better the analytical or formal orientation is, then, than the ‘funnel of causality’ approach of empiricists whose quest for the highest explained variance seldom produces more than a miscellaneous grab-bag of influences on the dependent phenomenon. Empirical work of that sort may have some limited utility in identifying possible causes, to be sure, but at some point the scholarly enterprise must move to the higher level of elaborating a clear logical structure among causal factors. Here, empirical success in accounting for observed phenomena cannot be the sole guide: the best theory is the one that provides the most accurate idea of what is actually going on in the real world, not the one with the best correlations.
Article
A key premise of partisan theories of congressional organization is that majority status confers substantial procedural advantages. in this article, Me take advantage of changes in party control of the House and Senate, such as that following the Republicans' historic victory in the midterm elections of 1994, to assess the value of majority status in terms of contributions from access-seeking political action committees (PACs). We estimate that majority status in the House was worth about $36,000 per member iii receipts from corporate and trade PACs circa 1994-even controlling for the usual factors cited in the literature as affecting members' ability to raise money (such as committee assigments and voting record). The value of majority status bt the Senate is even larger in absolute terms, although smaller in proportion to the total amount of money raised. Ou results show that majority status is a valuable asset, one worth considerable collective effort to attain.