veto players and referendums around the world
TRANSCRIPT
-
8/4/2019 Veto Players and Referendums Around the World
1/52
VETO PLAYERS AND REFERENDUMS AROUND
THE WORLD
Simon Hug and George Tsebelis
ABSTRACT
The literature on referendums comes to con icting assessments: for some
authors referendums are equivalent to direct democracy, for others, a poor
and unfounded substitute. In addition, existing classications use very diverse
criteria, and theoretical models lead to different results depending on whether
the underlying assumptions reect a single- or multi-dimensional policy space:
single-dimensional models lead to specic policy predictions, while multi-
dimensional models typically identify paradoxes connected with referendums.
We present a multi-dimensional model of referendums that leads to generaliza-
tions of the single-dimensional models. This multi-dimensional model makes
predictions about how various provisions for referendums affect policy
stability and the relationship between voter preferences and policy outcomes.
To assess the relevance of these predictions we present information about
referendum procedures all around the world on the basis of our model. Linking
this information with existing empirical work on the effects of referendumssuggests that our theoretical predictions are largely borne out.
KEY WORDS . institutional analysis . referendums . veto players
No act of [representatives] can be law, unless it has been ratied by the people in person;
and without that ratication nothing is a law. (Rousseau, 1947: 85)
[W]e are deluding ourselves if we consider the referendum and the popular initiative of
legislation as modern equivalents and substitutes of direct democracy. (Sartori, 1962: 256)
If the people speak in meaningless tongues, they cannot utter the law that makes them
free. (Riker, 1982: 239)
Introduction
Institutional provisions for referendums are increasingly being added to new
constitutions (e.g. in Eastern Europe, see Auer and Bu tzer, 2001). More and
more important questions are decided by referendum (e.g. recent changes of
Journal of Theoretical Politics 14(4): 465515 Copyright & 2002 Sage Publications
09516928[2002/10]14:4; 465515; 027705 London, Thousand Oaks, CA and New Delhi
Simon Hug acknowledges the partial nancial support of the Swiss National Science Founda-
tion (Grants 8210046545 and 50040487882/1) and the research assistance of Anke Tresch.
George Tsebelis acknowledges support by the Russell Sage Foundation.
http://www.sagepublications.com/http://www.sagepublications.com/ -
8/4/2019 Veto Players and Referendums Around the World
2/52
constitutions in Venezuela and Australia, ratication of international treaties
in the European Union, etc.). Increasingly scholars advocate greater use of
referendums (e.g. Budge, 1996; Zu rn, 1996; Abromeit, 1998; Ackerman,
1999; Schmitter, 2000; etc.). But in the theoretical literature Riker's (1982)
warning against populism together with recent work on `problems with
referendums' (e.g. Brams et al., 1997, 1998; Lacy and Niou, 2000), or the
`referendum paradox' (Nurmi, 1997a, b, 1998) still cast some doubt on the
usefulness of referendums. We argue that ultimately the difference in assess-
ments can be attributed to inadequate attention paid to the institutions
regulating referendums. More specically, scholars tend to neglect the
important question of who controls the referendum agenda. We argue that
this question is in fact separated into two parts: (1) who asks the question
and (2) who triggers a referendum.But the existing literature has a series of additional shortcomings. Some
authors classify referendums without paying attention to the strategic
choices of the different actors involved. The more theoretical studies come
to conicting conclusions depending on whether they assume that the under-
lying policy space is single-dimensional or has multiple dimensions. Single-
dimensional models lead invariably to sharp predictions of outcomes,
while multi-dimensional models underline possible paradoxes. In this respect
the referendum models mirror the rest of the spatial voting literature which
leads to median voter results in single-dimensional models and in chaos whenmore dimensions are introduced. This discrepancy is particularly damaging
to the study of referendums, because actual referendums sometimes address
a simple question, in which case a single-dimensional model is a reasonable
approximation, while at other times they are used to ratify whole constitu-
tions, in which case single-dimensional models are inadequate.
We present a multi-dimensional model of referendums on the basis of veto
player theory (Tsebelis, 1995a, 1999, 2000, 2002). According to this theory
veto players are actors whose agreement is necessary for a change in the
legislative status quo. Every government of a country represents a certainconguration of veto players.1 As the number of these actors and the ideo-
logical distances among them increase, a signicant change in the status
quo becomes more difcult (policy stability increases). The possibility of a
referendum introduces one additional veto player in each country: the popu-
lation. As a result, it moves policy outcomes closer to the preferences of
the median voter (if such a voter exists), but, provided that the remaining
466 JOURNAL OF THEORETICAL POLITICS 14(4)
1. See Tsebelis (2002) on how to identify and count veto-players as a function of the prevail-
ing institutions (presidential or parliamentary regimes, uni-cameral or multi-cameral legislatures,
required simple or qualied majorities) and the prevailing political game (number of parties in
government, ideological distances among them).
-
8/4/2019 Veto Players and Referendums Around the World
3/52
veto players keep their powers, it also makes signicant policy changes more
difcult.2 These are the similarities of all referendums. The differences among
them stem from institutional differences about who asks the question and
who triggers a referendum. We explain the signicance of these differences
and, on the basis of them, examine the institutional details of referendums
all around the world. In particular, we argue that if the same actor (whether
it is an existing veto player or not) controls both the formulation of the ques-
tion and the triggering of the referendum, other veto players lose their ability
to veto outcomes and hence the number of veto players actually decreases.
With our model we address the shortcomings in the different streams of
the referendum literature: unlike the conictual conclusions of single- and
multi-dimensional models in the theoretical literature, we analyze referen-
dums exactly the same way whether the underlying space has a single ormultiple dimensions; we distinguish referendums on the basis of the strategic
choices of the actors involved; we do so by paying attention to the institu-
tions regulating referendums; nally, we come to conclusions consistent
with the empirical evidence on the effects of referendums.
The paper is organized in ve sections. Section I reviews the different
approaches in the literature. We show that each one of these approaches
has serious shortcomings on the theoretical or the empirical level. Section
II introduces a model of referendums on the basis of veto player theory.
This model compares the institutions of a representative democracy and arepresentative democracy combined with referendums. Section III develops
the model further and points out the differences between referendums
depending on who is empowered to ask the question and who triggers a
referendum. On the basis of this classication, we discuss the consequences
of various types of referendums. Section IV uses the veto players model to
study the institutions of referendums around the world. In Section V we
show how well existing empirical evidence resonates with the theoretical
claims we advance. The conclusion suggests avenues for further research.
I. Different Literatures on Referendums
The literature on referendums can be subdivided in three groups. The rst
is empirical and often presents classications of the institutions allowing
citizens a direct say on policies. Each of these classications emphasizes
particular aspects of the referendum process, but most are disconnected
from any theoretical framework. The other two groups consist of theoretical
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 467
2. We will show later that our argument holds even in multi-dimensional spaces where such a
median rarely exists.
-
8/4/2019 Veto Players and Referendums Around the World
4/52
models that examine referendums as games among different actors. How-
ever, the rst of these two groups assumes that the questions asked can be
represented on a single dimension and as a result the outcomes of different
procedures can be calculated and identied, while in the second group the
multi-dimensionality assumption leads to a series of paradoxical and disturb-
ing results.
Classications
At the most basic level authors often distinguish between referendum and
initiative (e.g. Ranney, 1978: 69; Magleby, 1984; Cronin, 1989). The refer-
endum in this crude distinction is dened as a vote on a measure adopted
by parliament. The initiative, however, allows citizens to propose a ballotmeasure, which may be adopted in a popular vote. Often, authors add a
second criterion distinguishing whether the measure appearing on the
ballot is a constitutional amendment or a statute or law (e.g. Ranney,
1978: 69). These distinctions between two basic forms of referendums are
questioned in part by authors like Smith (1976), who rely on the functional
properties of the referendum device. Smith (1976) argues that referendums
can either be pro- or anti-hegemonic, and either controlled or uncontrolled.3
Combining these two criteria, Smith (1976) arrives at a fourfold classication
to which he adds an additional residual category, namely that of non-functional referendums, dened largely by their non-signicance. This classi-
cation, obviously, does not only rely on institutional properties of the
various referendum devices, but considers mostly contextual elements.
A classication based much more heavily on institutional characteristics
appears in Suksi (1993: 28f). This author proposes four dichotomous criteria
to distinguish among different institutional provisions allowing for referen-
dums. First he distinguishes between referendums that are mandatory and
facultative. The second criterion discriminates between decisive and con-
sultative. The third one distinguishes between pre-regulated and non-pre-regulated referendums, while the last one concerns whether citizens play an
active or passive role in launching a referendum. The combination of these
four dichotomies leads to a classication with (theoretically) 16 possible
types, several of which, however, are largely irrelevant since they do not
appear in reality.
In an attempt resembling Smith's (1976) classication, Hamon (1995) pro-
poses several classications emphasizing the control various actors have over
the referendum process. The most interesting aspect of his classication
effort is a simple table (Hamon, 1995: 24) showing possible combinations
468 JOURNAL OF THEORETICAL POLITICS 14(4)
3. Smith (1976) uses referendums as a generic term, also comprising initiatives.
-
8/4/2019 Veto Players and Referendums Around the World
5/52
between the author of a measure appearing on the ballot and the actor
triggering the referendum process. Both sets of actors comprise parliament,
government, the head of the state, parliamentary opposition and the citizens.
While some combinations do not exist in reality, the classication neverthe-
less goes a long way in emphasizing the difference between triggering a refer-
endum and formulating the question.
Uleri (1996) proposes a largely inductive classication relying on six
dichotomous criteria, which in part resembles the ones proposed by Suksi
(1993). Uleri (1996) arrives, however, only at a fourfold typology, compris-
ing decision-promoting initiative, decision-controlling initiative, decision-
promoting referendum and decision-controlling referendum. Among the
latter he distinguishes between rejective and abrogative votes. This classica-
tion attempt is picked up by Seta la (1999) and combined with some of Suksi's(1993) criteria. Mueller (1996: 177f ) distinguishes among four types of refer-
endums, namely the constitutionally mandated referendum, the government-
initiated referendum, the citizen initiated veto and the citizen initiative. Hug
(1999) derives an almost identical classication by employing two criteria
proposed by Suksi (1993), namely whether a referendum is required or not
and whether voters are actively involved or not in triggering a referendum.
For the non-required referendums, which are launched actively by the
citizens, however, he introduces as an additional distinction the criterion of
whether the ballot measure is proposed by the government or anotheractor. These latter two classications are almost identical, since both authors
explore the consequences of these four types of referendums in a strategic
context. In such a strategic context it is of prime importance to know who
triggers the referendum and who detains the agenda-setting power.
Theoretical One-dimensional Models
The last two classications also tie in very closely with existing theoreticalmodels of referendums. While early models in the public choice tradition
simply presumed that referendums would correct any type of rent-seeking
behavior of government (see Mueller [1979] for a discussion of this litera-
ture), Romer and Rosenthal (1978, 1979) studying different types of
school-bond referendums and demonstrated how required referendums
may lead to policy outcomes quite far away from the one preferred by voters.
Extending Romer and Rosenthal's (1978, 1979) work, Steunenberg (1992)
proposes a generalization of these insights over a whole set of possible insti-
tutions for policy-making in a single dimension. Gerber (1996) presents amodel more closely reecting the initiative process. However, this earlier
work is limited in two ways, namely by assuming rst complete and perfect
information for all actors involved in the decision-making process and,
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 469
-
8/4/2019 Veto Players and Referendums Around the World
6/52
second, that the policy to be decided can be represented in a one-dimensional
policy space.
Relying on more recent work in the agenda-setting literature, several
authors relaxed the rst limitation by assuming some informational asym-
metry among the actors involved. Lupia (1992, 1994) focuses on the limited
information voters may have about ballot proposals and nds that endorse-
ments by interest groups may provide helpful decision cues.4 Matsusaka and
McCarty (2001) propose a model of the initiative process where both the
legislature and an interest group fail to know the voters' exact preferences.
They nd that for most preferences the initiative process allows for policies
more closely aligned with the voters' wishes. A similar model of the initiative
process appears in Besley and Coate (2001). While they consider two policy
issues that have to be decided, in equilibrium only one of them appears on theballot, while the other is resolved in the legislature. Hence, despite the two
issues, voting on the ballot proposal is represented on a single dimension,
and the two authors also nd that the voters' wishes are more closely
reected in policy outcomes if initiatives are possible. Moser's (1996)
incomplete information model of the popular referendum process comes to
a similar conclusion. Hug (1999) explores the effect of incomplete informa-
tion for four types of referendums and nds that the effects of these institu-
tions differ. In general, however, for most preference proles voters (i.e. the
median voter) are better off having referendums at their disposal which theymay trigger themselves or which are required. While these theoretical models
relax the complete information assumption of previous models, they do not
question the assumption of a single-dimensional space.
Theoretical Multi-dimensional Models
The multi-dimensionality of policy decisions in referendums appeared
much more strongly in work relying on social-choice-theoretic approaches.
McKelvey (1976) and Schoeld (1978) demonstrated the importance ofagenda-setting. In their analysis the agenda-setter could produce literally
any outcome by sequencing appropriately the questions and exploiting the
different preferences among the different individual actors. Based on their
ndings Riker (1982) argued strongly against the existence of a Rousseauean
`general will' in multi-dimensional policy spaces.
Related to these theoretical models is recent work on some paradoxes
which may occur in referendum voting, because either policies are multi-
dimensional or voters vote on various non-separable proposals at the same
470 JOURNAL OF THEORETICAL POLITICS 14(4)
4. Bowler and Donovan (1998) provide persuasive empirical tests of these implications as well
as others in the context of referendum voting in the United States.
-
8/4/2019 Veto Players and Referendums Around the World
7/52
time. Brams et al. (1997, 1998) and Saari and Sieberg (2001) focus on the
problem of multiple measures appearing on the same ballot. While they
are concerned with the fact that possibly no voter has voted for the winning
`combination', Lacy and Niou (2000) explore this problem from the angle of
non-separable preferences. Provided that various ballot measures are related,
voters may have preferences over a particular ballot measure which are
dependent on the outcome of another ballot measure. But since voting
occurs simultaneously over all ballot measures at the same time, voters
with non-separable preferences cannot condition their vote on the outcome
of another ballot measure. Lacy and Niou (2000) suggest that referendums
are not very good at resolving the problems stemming from non-separable
preferences and suggest that legislatures deal more effectively with these
issues. Hence, they advocate the use of normal legislative procedures forissues where voters may have non-separable preferences. Brams et al.
(1997, 1998), however, suggest using approval voting to overcome the
problems of multiple ballot measures.
Nurmi's (1997a, b, 1998) work relates much more strongly with social
choice theory and suggests a series of paradoxes that may appear in referen-
dums. While this theoretical and empirical work related to social choice
emphasizes important problems of referendums, it is useful to take a step
back. Work in this tradition assumes well-behaved preferences over indi-
vidual proposals while emphasizing that these preferences may be non-separable from those on other proposals. In addition, this work, with the
exception of Nurmi (1997a, b, 1998) does not consider how referendums
interact with representative democracy. In conclusion, the disorderly image
of positive and negative results nding also reection in our quotes at the
beginning of our paper persists when one focuses on the accounts in
the literature. Multiple criteria and different classications characterize the
more empirically oriented literature, while specic outcomes or paradoxes
dominate the more theoretical approaches. Our goal is to provide a unied
framework for the study of referendums and to demonstrate two points.First we show that the actual provisions prevailing around the world can
be understood on the basis of our framework and, second, we demonstrate
that the results generated by empirical analyses are congruent with the expec-
tations generated by our theory.
II. Veto players, Direct and Mediated Democracy
Our analysis of the institutional frameworks of referendums is based on theconcept of `veto players' and expands on the work of Tsebelis (1995a, b,
1999, 2000, 2002). According to Tsebelis (1995a) a veto player is an indi-
vidual or collective player whose agreement is required for a change of the
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 471
-
8/4/2019 Veto Players and Referendums Around the World
8/52
status quo. This denition can be applied to all countries and allows for iden-
tifying the actors involved in policy-making whether they are individual (like
Presidents with the power to veto legislation) or collective (governments, one
or both chambers of parliament, parties) actors. Tsebelis (1995b) demon-
strates that in countries under a parliamentary regime the actual veto players
are most of the time the parties in government.5 Tsebelis demonstrates both
theoretically (Tsebelis, 1995a, 2002) and empirically (1999, 2002) that the
number of veto players and the ideological distances among them make
changes to the status quo more difcult. In addition, he points out (2000)
how the interactions in referendums can be understood on the basis of the
veto player framework.
What difference does it make if outcomes are selected directly by the
people or indirectly by the people's representatives in parliament? By deni-tion outcomes selected by parliament will be preferred over the status quo by
a majority in parliament, while outcomes selected by a referendum will be
preferred by a majority of the voters. In a single dimension the outcome
will be determined by the corresponding median voter, but in multiple
dimensions such a median voter very rarely exists. The number of policy
dimensions involved in a referendum is an open question. Sometimes multi-
ple issues are lumped together; at other times efforts are made to separate
issues and decide them one at a time. For example, on the one hand, refer-
endums are sometimes used to approve (or disapprove) whole constitutions;on the other hand, the Italian Constitutional Court decided to exclude a
popular proposal on the basis that it contained `such a plurality of hetero-
geneous demands that there was a lack of a rational, unitary matrix that
would bring it under the logic of Article 75 of the Constitution' (quoted in
Bogdanor, 1994: 634).
This section will rst make the argument that in our model (in sharp con-
trast with the literature that we reviewed) the number of underlying dimen-
sions makes very little difference. The second issue that we will address is
that the preferences of such an `as if' median voter may be signicantlydifferent from the policy selected by existing veto players.
Median and `As If' Median Voter Preferences in Referendums
If we consider a group of people voting by majority rule whether to accept
a proposal or reject it (and keep the existing status quo (SQ)), the set of
472 JOURNAL OF THEORETICAL POLITICS 14(4)
5. However, there are exceptions, for example if the President of the Republic has veto power
over legislation (Portugal) or ordinances (France), or in bicameral systems, if the upper cham-
ber's approval is required and a different majority controls this chamber (Germany some of
the time).
-
8/4/2019 Veto Players and Referendums Around the World
9/52
outcomes that can defeat the SQ form the win-set of the status quo: W(SQ).
Social choice literature (Ferejohn et al., 1984) has identied an area within
which W(SQ) is located. This area is a circle centrally located within the
group of voters. More precisely, Ferejohn et al. make the assumption that
voters' preferences are represented by points in space and each voter is
indifferent between two alternatives equal distance from their ideal point
(Euclidean preferences); they draw all the `median' lines within the voter
population.6 They then consider the smallest circle intersecting all these
lines and call it the `yolk'. If the yolk has center Y and radius r, and the
segment SQY has length d, they show that W(SQ) is included in a circle
(Y, d 2r), and includes a circle (Y, d 2r).7
We will take this nding as the departing point in our analysis.8 As a result,
the boundaries of the win-set of the status quo are located between twocircles; both of them with center Y, one with radius (d 2r) and the other
with radius (d 2r). It has also been shown (with computer simulations)
that on average the size of the radius of the yolk (r) decreases as the
number of voters increases (Koehler, 1990).9 Consequently, for the millions
of people who are the potential participants in a referendum, in most
countries or states r is (most of the time) exceptionally small. As a result,
the win-set of the status quo is contained between two circles that differ
little from each other: 4r, when r becomes smaller and smaller.
What the previous two paragraphs indicate is that for a large population,the median voter may not exist but all median lines pass through a very small
area (of radius r), so an `as if' median can be very well approximated by the
center Yof the yolk of the population. In addition, the win-set of the status
quo for such a large population is also very well approximated by a circle
with radius d. In other words, the multiplicity of voters simplies rather
than complicates the problem of identication of the median voter and the
win-set of the status quo.
Figure 1 provides a visual representation of the argument. The yolk of
the population is very small and has center Y. The win-set of a point thathas distance dfrom Yis the shaded area in the gure, and is located between
the two circles with radii (d 2r) and (d 2r), so it can be approximated by
the circle (Y, d).
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 473
6. Median lines have on both sides of them majorities of voters (including the points on
them).
7. Ifd>
2r
; otherwise such a circle does not exist.8. The interested reader can nd a simple proof of these statements in Miller et al. (1989).
9. We underline `on the average' because one can nd counterexamples. For example, if the
voting population has only three distinct preferences, then there is no difference between such a
population and a three-member voting body.
-
8/4/2019 Veto Players and Referendums Around the World
10/52
Representative Democracy
Consider a country with ve parliamentary parties, A, B, C, D and E, three of
which are required to form a parliamentary majority (a simple example of
that would be if each one of them had 20 percent of the seats). Where
would a parliamentary decision be located? If we know nothing about the
parliament's decision-making except that it requires a simple majority
(think of a decision taken in a presidential system where coalitions depend
on the issue under consideration), then we have to locate the intersections
of any three of the circles with centers A, B, C, D and E. We have shadedthese areas in Figure 2. If we know some additional information about
parliament's decision-making we can incorporate it in the calculations, and
identify the win-set of the status quo more accurately. For example, if we
know the parties that form the government of a parliamentary system, we
will identify the win-set of the status quo by considering each one of them
as a veto player and nding the intersection of their win-sets. Figure 2 indi-
cates with dark shade the area where the outcome will be located if the parties
forming the government are A, B and C in a parliamentary system govern-
ment (where coalitions are, by denition, stable). In the remainder of thepaper, when we use the term `parliamentary' we will imply a system with
stable majorities, while when we use the term `presidential' we will imply
474 JOURNAL OF THEORETICAL POLITICS 14(4)
Figure 1. Win-set of a Large Group of Voters
-
8/4/2019 Veto Players and Referendums Around the World
11/52
the possibility of different coalitions depending on the issue. What happens if
representative democracy and referendums are combined?
Representative Democracy and Referendums
Denmark provides some interesting examples of the differences between
the outcomes from representative democracy and referendums. As Vernon
Bogdanor (1994: 72) puts it:
It may seem a paradox that the Single European Act, which could not have gained a
majority in the Folketing, received a majority in the country, while Maastricht, which
enjoyed the support of parties with 80 percent of the seats in the Folketing, was rejectedby the voters in 1992.
Figure 3 helps us think the potential paradox through.
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 475
Figure 2. Win-set of Status Quo in Representative Democracy
-
8/4/2019 Veto Players and Referendums Around the World
12/52
There is no reason to believe that the two processes (representative democ-
racy and referendum) will lead to the same outcome. In fact, we do not even
know that the preferences of the public will be centrally located within the
preferences of the parties. Even the most pure proportional representation
systems like Israel or The Netherlands cannot guarantee representation for
minorities of 0.5 percent, for example. Systems with higher thresholds like
Sweden's 4 percent or Germany's 5 percent exclude many more. In addition,
the existence of electoral districts may create serious differences between the
preferences of the voters and their representatives.10 Finally, parliamentary
institutions (like committees, qualied majorities) may also create discrepan-
cies between popular preferences and parliamentary outcomes (see Tsebelis,
2002).
476 JOURNAL OF THEORETICAL POLITICS 14(4)
Figure 3. Difference of Results Between Direct and Representative Democracy
10. The literature on referendum paradox (Nurmi, 1998: 3367) shows one mechanism
generating such a discrepancy.
-
8/4/2019 Veto Players and Referendums Around the World
13/52
In Figure 3 we present how a referendum interacts with the parliamentary
arena. In order to capture the differences in preferences between the people
and their representatives we represent the center of the yolk Y H generated
by the preferences of the people not in the center of the parties' preferences.
On the basis of the analysis for median voter preferences, the outcome of a
referendum will be located inside the hatched area called W H (SQ). On the
basis of the analysis for representative democracies, the outcome of a presi-
dential system would be anywhere in the area W(SQ), while the outcome of a
parliamentary system with parties A, B and C in government would be
located in the very dark area. One can see that the possible solutions under
representative democracy and referendums have several points in common,
but the preferences of the population and the political parties do not neces-
sarily coincide. This is only part of the story because there is no guaranteethat the coalition prevailing among the voters would be politically the
same as the coalition prevailing in parliament. For example, the parlia-
mentary government A, B, C would produce an outcome located inside the
very dark area, while the outcome of a referendum could be anywhere
inside the hatched area in Figure 3.
If a parliamentary decision has to be ratied by the population (as is
frequently the case in constitutional matters) then the outcome has to be
located in the intersection of the parliamentary and the popular win-sets.
In other words, referendums create one additional veto player in thedecision-making process: the people. There are two results from this introduc-
tion of a new veto player: rst, it becomes more difcult to change the status
quo (Tsebelis, 1995a). Second, the nal outcomes will approximate the prefer-
ences of the `as if' median voter better when the possibility of a referendum
exists (whether the actual decision is made by a referendum or not).11 But
the institutions that regulate referendums affect both of these results.
III. Institutions Regulating Referendums
From the perspective of the veto player theory two main criteria are funda-
mental. First, we have to know who may trigger the process of a referendum.
Second, we need to know who is able to frame the question of the popular
vote. These two crucial questions underlie, in part, some of the classications
discussed earlier, but they are seldom pursued to their ultimate conclusions.
Stating these questions clearly and using them as the main criteria for a
classication of different institutional provisions for referendums has the
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 477
11. Gerber and Hug (2001) discuss in detail these direct and indirect policy effects of institu-
tional provisions allowing for referendums.
-
8/4/2019 Veto Players and Referendums Around the World
14/52
additional advantage of allowing theoretical claims on referendums to be
easily tied into insights gained about representative democracy from veto
player theory.
Figure 4 employs these two basic questions to derive a fourfold classica-
tion which comes very close to the ones proposed by Mueller (1996) and Hug
(1999). There are some slight differences. Mueller and Hug dene distinctions
among referendums on the basis of what government and opposition do
though these concepts may not be well dened in presidential systems.
Here we make distinctions on the basis of the behavior of actors identiable
in every political system: veto players and non-veto players.12 This allows us
to make a much clearer distinction with respect to the two main criteria. How
these criteria are used is best presented in a treelike fashion in Figure 4. First,
478 JOURNAL OF THEORETICAL POLITICS 14(4)
Figure 4. Questions Dening Different Categories of Referendums
12. There is an additional minor distinction, namely that even in parliamentary systems
`government' and `veto-players' are not always identical (see footnote 6).
-
8/4/2019 Veto Players and Referendums Around the World
15/52
we have to distinguish between referendums that are required and those that
are not. Obviously, in the former case the question of who triggers the refer-
endum process is irrelevant. Non-required referendums come in different
forms and shapes and distinguish themselves by the two criteria. With respect
to the question of who decides to trigger a referendum, two possibilities have
to be taken into consideration. First, a veto player in the normal legislative
arena might trigger a referendum. Second, an actor who is not a veto
player in the representative system might trigger a referendum: popular
vetoes and initiatives are prime examples. When a non-veto player triggers
a referendum the crucial question becomes who asks the question. In the
Italian case the question is asked by parliament, since referendums may
only strike existing laws (or parts thereof) from the books. In other cases,
an actor who does not have veto player status in the legislative arena mayask the question. Popular initiatives, for instance in the States of the USA
or in Switzerland, are the main examples. This much more ne-grained
classication into four major types allows a very detailed analysis of the
policy consequences of the various types of referendums. We discuss these
policy effects for each referendum separately.
Required Referendums
At the most basic level required referendums introduce an additional vetoplayer. Apart from the veto players whose support is normally necessary
to change policies, required referendums add the hurdle of the support of
the voters. For policies that are the object of required referendums, e.g. con-
stitutional amendments in many countries or changes in the voting age in
Denmark (see Table 1), the status quo can only be changed if the change is
preferred by the new veto player, namely the voters. Thus, any new policy
has to belong to the win-set of the status quo as dened by the voters' prefer-
ences. In most cases this win-set corresponds to the set of policies that is pre-
ferred by a simple majority of the voters to the status quo. Often, however,provisions for required referendums specify qualied majorities.13 The
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 479
13. For instance in the federal system of Switzerland constitutional changes require a major-
ity not only among the voters, but also among the cantons which make up the federation. In
Italy, however, a referendum outcome is only valid if turnout exceeds 50 percent. In Denmark,
a rejection of a bill transferring some aspects of national sovereignty to an international organi-
zation is only valid if at least 30 percent of the voters having participated at the last election reject
the bill. If less than this minimal percentage of votes against a proposal it is deemed to beaccepted by the voters. This almost resembles the rules under which Napoleon held the rst
national referendum in Switzerland on adopting a new constitution for the Helvetic Republic.
Although roughly 56 percent of the voters rejected the new constitution Napoleon declared it
adopted, since the citizens having failed to vote were obviously in favor (Mo ckli, 1994: 60).
-
8/4/2019 Veto Players and Referendums Around the World
16/52
model can easily accommodate such differences in majority requirements
(e.g. Tsebelis, 2002). They simply affect the nature of the veto player intro-
duced by referendums. In countries with no turnout criterion the veto
player is simply formed by the collectivity of participating voters. In
countries with a turnout requirement, some form of qualied majority of
voters makes the decision. Despite these considerable variations we limit our-
selves in this paper to explore the consequences of referendums of various
types under the assumption of simple majority requirements. Under this
assumption our model predicts that policies in issue areas subject to required
referendums will belong to the intersection of the parliamentary win-set
(W(SQ) in Figure 3) and the voters' win-set (W H(SQ) in Figure 3). Figure 3
illustrates that a majority of voters will not be worse off due to required
referendums, because required referendums rule out elements of the parlia-mentary win-set that are worse than the status quo for a majority of
voters.
Veto player Referendums
From the previous discussion it becomes clear that only under extremely rare
conditions would the possible outcomes of the two processes (the win-sets of
representative democracy and referendums) be identical. In terms of actualoutcomes, they can be identical even when the win-sets differ, or different
even when the win-sets are the same. This obviously gives considerable
leeway to the veto player who may trigger a referendum, and even more so
if this veto player may also formulate the referendum question. We rst
consider veto player referendums in which a veto player may trigger a refer-
endum, but the normal legislative procedures leads to the formulation of the
ballot question. Then we move on to the case where the same veto player
formulates the question and triggers the referendum.
Let us focus on Figure 3 and see under what conditions different vetoplayers actually would call for a referendum. In a purely representative
system the policy outcome would belong to the win-set of the status quo
W(SQ). But any policy not belonging to W H (SQ), the voters' win-set of the
status quo is open for a referendum challenge. In order to simplify our
calculations, let us assume that a referendum has no political costs for the
agenda-setter. Obviously, this is an incorrect assumption, but one can
address it easily by adding these costs in the calculations. For instance, if
the legislative process led to an outcome in the small petal of the win-set to
the lower right, most likely B, but certainly C would trigger a referendumwhich would result in the rejection of the new policy by the voters. Thus if
B and C are veto players and can trigger a referendum, the referendum
threat will lead to the adoption of a policy in the intersection of the two
480 JOURNAL OF THEORETICAL POLITICS 14(4)
-
8/4/2019 Veto Players and Referendums Around the World
17/52
win-sets W(SQ) and W H (SQ).14 For the case in which the veto player triggers
a referendum and asks the question we will consider the two different cases of
`parliamentary' and `presidential' systems (stable coalition of parties A, B
and C in the rst case and any coalition of three of A, B, C, D and E in
the second). In each of these cases we consider two possible agenda-setters:
party A and party E (the rst is part of the parliamentary system govern-
ment, the second is not).15
Under complete information the referendum agenda-setter is guaranteed
to get his/her most preferred point from the popular win-set of the status
quo (W H (SQ) in the picture).16 Given that both A and E are located outside
W H (SQ) they can achieve the points AH and EH respectively when they control
the referendum agenda. The question is: Can indirect democracy offer to the
referendum agenda-setters a more attractive alternative? In order to answerthis question we have to calculate the win-set of these two points W(A H) (see
Figure 5) and W(E H) (see Figure 6).
Figure 5 presents exactly the same conguration of players as Figure 3 and
identies the point AH, which is the best outcome the referendum agenda-
setter can achieve (AH is the intersection of the line AY H) with the circle
(center Y H, radius jYH SQj). Figure 5 also identies the win-set of AH instead
of W(SQ), since player A can introduce a referendum and obtain AH as the
outcome. Out of this win-set A will consider only the points included in
the circle (A, AAH
), and trigger a referendum for any point further awaythan AH. There is only one possible coalition that can approve points inside
the (A, AA') circle: A, D and E. Consequently A has to select this coalition
in order to get an outcome which is preferred over A H (preferably A?). In our
idealized `presidential' system this is what will happen. In the case of a
`parliamentary' system with A, B and C in government the situation is
more complicated. Note that there is no point that all three parties A, B
and C prefer to AH because AH is in the unanimity core of A, B and C.
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 481
14. If B and C are not veto-players who can trigger a referendum, for instance if A, D, E form
a coalition, then the outcome in the small petal of the win-set to the lower right will not be chal-
lenged in a referendum. Since this scenario is only possible in a popular veto, we will cover it
later.
15. We consider that agenda-setting belongs to single parties, because then it is easy to iden-
tify the proposal that they will make. If several parties (say `the government') control the refer-
endum agenda they will have to negotiate among them what proposal they will make. We have
nothing to say about these negotiations, except that feasible outcomes have to be within W H(SQ).
Note also that E is not a veto-player. We include this counterfactual case (see Figure 6) in order
to show the underlying strategic calculations under all possible congurations, whether they existin actual political systems or not.
16. In the absence of complete information such precise calculations are impossible. Actors
will select points that they think are inside WH(SQ) but they may be wrong in which case the
voters will reject the proposal.
-
8/4/2019 Veto Players and Referendums Around the World
18/52
A has to choose between keeping the government in place or leading a
government to resign. Similarly, parties B and C may offer to approve out-
come AH and avoid a referendum, or they might prefer to delegate their dis-
agreement to a referendum. These calculations lead to three possible
outcomes: (1) government ABC remains in power and adopts AH without a
referendum; (2) government ABC remains in power and AH
is adopted byreferendum; (3) the government resigns, and is replaced by another coalition
which selects a feasible point from W(AH).
482 JOURNAL OF THEORETICAL POLITICS 14(4)
Figure 5. Possible Outcomes When A Controls the Referendum Agenda
-
8/4/2019 Veto Players and Referendums Around the World
19/52
Figure 6 presents exactly the same conguration of players as before, but
assumes that player E can introduce a referendum, so identies the win-set of
point E H instead of SQ. Out of this win-set only the points included in the
circle (E, EE H) can be considered, because E would prefer to trigger a refer-
endum than to accept a point further away than E H. There are three possible
coalitions that can approve points inside the (E, EE H) circle: (ABE), (ADE),
and (CDE). Consequently E has to select one of the available coalitions. In
the case that any coalition is possible (the `presidential' system), E will select
his own ideal point supported by (ADE). In the counterfactual case of a
`parliamentary' system with (ABC) in government the situation would be
more complicated. E could use his/her advantage of referendum agenda-
setting to try to negotiate a different government: indeed, players A, D and E
may prefer a new coalition government. If the parties in government want tostick together, E will trigger a referendum and the government will lose.
In all these calculations some previous existing veto players lose their
ability to participate in political decision-making. Instead of calculating
W(SQ) we were basing our calculations on what the referendum agenda-
setter could obtain (points AH or E H). In addition, we saw that a presidential
system, where parties can shift coalitions on the basis of the subject matter
under consideration, was a more exible system than a parliamentary one,
where the existing government coalition was unable to adapt to the new
policy environment generated by the referendum not only when a non-vetoplayer controlled the agenda (case of E being the agenda-setter), but also
when agenda-setting belonged to an existing veto player (the case of A). A
parliamentary system can produce similar outcomes by delegating a political
issue to a referendum and leaving it outside the political conict of the main
parties. For example, in the UK the referendum on participation in the EU
had this special treatment because both parties were divided and could not
handle the issue without serious damage to their unity (Bogdanor, 1994).
Given these calculations, strategically thinking parties in the legislature
(particularly if, for some reason, they want to avoid a referendum) canassure the referendum agenda-setter that they will do anything in their
power to make the legislative process end up in an area that is at least as
good for him/her as the result of a referendum. These mental experiments
lead to the following conclusions. First, the position of referendum
agenda-setter translates into signicant policy advantages: if a veto player
controls the referendum agenda, s/he cancels other veto players as such.
This is a very different analysis than the one presented by referendum advo-
cates who consider referendums the expression of the will of the people (see
earlier quote by Rousseau). Second (and this is a consequence of the rst),the legislative outcomes of representative democracy are altered if referen-
dums are possible (Gerber and Hug, 2001).
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 483
-
8/4/2019 Veto Players and Referendums Around the World
20/52
484 JOURNAL OF THEORETICAL POLITICS 14(4)
Figure 6. Possible Outcomes When E Controls the Referendum Agenda
-
8/4/2019 Veto Players and Referendums Around the World
21/52
Popular Veto
If popular vetoes are possible, the outcome of the legislative process has to
lead to policies contained in the win-set of the status quo dened by the
voters. If this is not the case, an actor may trigger a referendum, providedthat s/he prefers the status quo to the policy adopted. In Figure 3 this
would suggest that in the parliamentary arena a policy is adopted which
lies outside W H(SQ), the set of policies the voters prefer to the status quo.
If an actor prefers the status quo to the proposed policy and has the capacity
to trigger a referendum, it can ensure that the outcome will remain the status
quo.
An illustrative example in this respect is the rst Danish Maastricht refer-
endum. While a large majority accepted the Maastricht Treaty in the ratica-
tion debates, a minority exceeding one-sixth of the members of parliamentvoted against the Maastricht Treaty and forced a referendum.17 This minor-
ity in parliament evidently preferred the status quo to the Maastricht Treaty.
As the rst referendum showed, a majority of the Danish voters also pre-
ferred the status quo to the Maastricht Treaty, suggesting that this treaty
was not in the win-set of the status quo.
Hence, popular vetoes allow non-veto players to force a return to the
status quo ante. How this status quo ante is dened may differ quite con-
siderably. For instance, in the rst Danish referendum on the Maastricht
Treaty referred to earlier the status quo corresponded simply to the arrange-ments under the Single European Act (SEA). In other cases the status quo
ante is more complex. For instance, the Italian referendum that struck any
reference to proportional representation from the electoral law for the
Senate (Newell and Bull, 1993), led to a situation where the senators had
to be elected by majority rule. In this case the status quo ante is quite differ-
ent from the one enforced by the Danish voters.
Popular InitiativeSo far, we have dealt with referendums where the question was asked by
existing veto players. Now we focus on referendums delegating agenda-
setting powers to the winner of a competitive process. If different groups can
become agenda-setters (both asking the question and triggering) of a referen-
dum by winning the right to present their question to the electorate (signature
collection) the legislative outcome will depend on how competitive the
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 485
17. This example also illustrates that members of parliaments may launch a popular veto
simply by voting against a bill. Provided that a rejection by a qualied minority automatically
triggers a referendum, the mechanisms of such votes correspond perfectly to popular vetoes
(Table 2).
-
8/4/2019 Veto Players and Referendums Around the World
22/52
selection process is. If all potential players are included in the selection pro-
cess the only way that one can select proposals that will not only defeat the
status quo but other proposals as well is to make proposals that are sup-
ported by a majority, which means that the process will converge towards
the preferences of the `median voter'.
If some of the potential agenda-setters (particularly the ones with prefer-
ences similar to the `as if' median voter) are excluded from the process,
then the remaining ones may be more extreme and the legislative outcome
may be further away from the preferences of this voter. As a consequence
of this analysis, we have to focus on the process of selecting the agenda-
setter and assess how competitive it is. This selection obviously depends on
the criteria for the qualication of a ballot measure. These criteria comprise
the number of signatures of citizens required to qualify the ballot, how muchtime the qualication process can take, etc. All these elements thus create
hurdles a group has to cross. Gerber (1999) links this with the notion that
popular initiatives are launched either by broadly based citizen groups or
special interest groups. She argues that given the hurdles a group has to
cross to qualify a proposal, broadly based citizen groups, which will collect
signatures in part or mostly through volunteers, will mostly submit moderate
proposals. This simply because only moderate proposals, i.e. proposals close
to the center of the yolk, will nd sufcient support in the population to allow
volunteers to collect enough signatures. Special interest groups, which oftenrely on paid signature collectors, may make either moderate or more extreme
proposals, depending on how strongly they feel about a particular policy.
Consequently, if, for example, what is required is signature selection by
volunteers, then, demands that are supported by a majority of the population
are likely to get the volunteers necessary for their placement on the ballot,
and initiatives that do not have enough volunteers are unlikely to be sup-
ported by a majority. Consequently, such a process is a competitive one
and one can expect that the outcome will be located close to the preferences
of the median voter.If, however, what is requested for an issue to be placed on the ballot is
signature selection by remunerated professionals, it is possible that organized
groups with ideal points far away from the median voter are also able to
participate, in which case the selection process for agenda-setting may trans-
late into outcomes away from the preferences of the median voter. In all
cases, the selected outcome has to be closer to the preferences of the
median voter than the status quo.18 So, again, despite the fact that the `as
486 JOURNAL OF THEORETICAL POLITICS 14(4)
18. We have to repeat here that the radius of the yolk is assumed 0, otherwise the selection of
a point that is further from the center of the yolk than the status quo by up to 2r could not be
excluded.
-
8/4/2019 Veto Players and Referendums Around the World
23/52
if' median voter makes the nal decision, the result depends crucially on the
preferences of the agenda-setter.
Institutions for Referendums and Their Policy Consequences
Our model makes predictions in two related domains. First, it shows how
referendums, by introducing a new veto player and possibly eliminating
existing ones, affect the potential for policy change. Second, it also shows
that referendums affect considerably policy outcomes, namely in such a
fashion that voters (i.e. the median voter) are never worse off having provi-
sions for referendums. Both predictions relate closely to the concerns in the
general literature on referendums. But our predictions are of a much sharpernature, provided we have information on the institutional rules regulating
referendums and, even more importantly, can determine clearly the identity
of the actors able to trigger a referendum and to ask the question. Thus,
before discussing in more detail our predictions and how they relate to
empirical studies on referendums, we provide information on provisions
for referendums in constitutions of countries around the world.
While various studies provide partial information about the institutions
allowing for referendums, none relate this information with the notion of
veto player which in our analysis is crucial for understanding the relationshipbetween referendums and representative democracy. For instance, Suksi
(1993) presents information for most countries on whether their constitu-
tions have provisions for referendums and classies them according to his
dichotomous criteria. Similarly, the Research and Documentation Centre
on Direct Democracy (http://c2d.unige.ch) provides information on the insti-
tutions present in the countries around the world. But both these sources do
not directly provide information on whether or not veto players trigger a
referendum and formulate the question, so they do not provide the crucial
information identied in Figure 4. We provide this information in Tables 1and 2 in the Appendix. In collecting this information we relied heavily on
Suksi's (1993) work, but updated it considerably, adapting it to the needs
of our classication presented in Figure 4. A major criterion in this classi-
cation distinguishes between required and non-required referendums. In
Table 1 we list all countries that have provisions for required referendums
and explain the issues subject to such referendums as well as the date of
the constitution consulted.19
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 487
19. Suksi (1993: 138, 142) nds that in 56 out of his 160 analyzed constitutions there appear
references to referendums on constitutional amendments. Among them 32 explicitly refer to
required referendums for constitutional changes. Six other constitutions introduce a required
referendum for international treaties, namely Congo, Gabon, Mali, Senegal, Singapore, and
-
8/4/2019 Veto Players and Referendums Around the World
24/52
Overall we identify 54 constitutions containing provisions for required
referendums. The number of constitutions allowing for non-required referen-
dums (Table 2) is slightly smaller (47). For all of these 47 countries we pro-
vide detailed information on who may trigger a referendum and the identity
of the actors who ask the question.20 We use this information to determine
whether a veto player or a non-veto player may trigger a referendum and for-
mulate the question to be submitted to the voters. As our theoretical model
clearly indicates, it is this information which is crucial to assess the effect of
institutions allowing for referendums.21 Table 2 illustrates the wide variety
that exists in the institutional provisions allowing for non-required referen-
dums. Predominantly, we nd that existing veto players may trigger a refer-
endum and retain absolute control over the agenda. Only a few constitutions
contain provisions for popular vetoes or popular initiatives.The information contained in Tables 1 and 2, combined with the insights
from our model allow us to qualify claims prevalent in the literature on refer-
endums which discuss the policy and stability-inducing effects of referen-
dums. With respect to the stability-inducing effects of referendums, the
literature is replete with examples and discussions. Neidhart (1970), for
instance, discusses in detail the case of Switzerland, for which he argues
that the popular veto has made policy changes very difcult. Cronin (1989:
222), however, suggests that `direct democracy devices . . . have been an occa-
sional remedy, and generally a moderate remedy, for legislative lethargy . . .'.Reviewing the evidence of whether referendums have a conservative bias,
Gallagher (1996: 237) comes to the conclusion that there is no clear evidence
in support of this view nor its counterpart.
These differences in assessment relate directly to the ndings based on our
model. Generally speaking, referendums introduce an additional veto player
into the political game and thus diminish the potential for policy change.
While this stability enhancing effect is present in all types of referendums
488 JOURNAL OF THEORETICAL POLITICS 14(4)
Zaire (Suksi, 1993: 143). For each of these countries we consulted the most recent constitution
and report the detailed procedures that lead to a required referendum in Table 1. Given that
countries in Eastern and Central Europe (e.g. Lesage, 1995; Auer and Bu tzer, 2001) adopted
new constitutions in the 1990s, often with provisions for referendums, we also systematically
consulted these constitutions and list provisions for required referendums in these countries in
Table 1.
20. Again, we used Suksi's (1993) work to identify all countries that had non-required refer-
endums according to the constitutions he consulted. In addition, we checked all constitutions of
countries from Eastern and Central Europe to determine whether they had provisions for non-
required referendums. For all these countries we determined the relevant articles in the constitu-tion to determine the exact procedures for holding such referendums.
21. While several other sources also provide lists of provisions allowing for referendums,
none of them identies the actor triggering a referendum and formulating the question for the
voters in terms of veto-players.
-
8/4/2019 Veto Players and Referendums Around the World
25/52
of our classication (Figure 4), our model also identies the cases that lead to
the elimination of veto players: if the same player (whether an existing veto
player (veto player referendums) or not (popular initiatives)) controls both
triggering and asking the question. First, in a veto player referendum, if a
subset of the normal legislative veto players can both trigger a referendum
and ask the question, then this cancels out the remaining veto players in
the legislative arena (other parties in government). An example for this
type of referendum is the provision of Belarus' constitution allowing the
President to trigger a referendum and to ask any question. The effect of
such powers clearly transpired in the referendums in 1996 in which President
Lukashenka broke a stalemate with parliament (Stone, 1997).22 Proposals
for increasing the president's powers would never have been passed in parlia-
ment. But Lukashenka appealed over the heads of the parliament to theBelarusian voters and changed policy in an area where representative democ-
racy would not have produced any change.
The other type of referendum eliminating existing veto players, and thus
not necessarily decreasing the potential for policy change, is the popular
initiative. Again, this comes about because the power to set the agenda in
these referendums is vested in an actor who is not a veto player in the
normal legislative game. If citizens can submit their own policy proposals
and trigger a referendum, they cancel out the powers of existing veto players.
Thus, countries having provisions allowing for the popular initiative may seethe potential for policy change actually increase. Given that this increase
comes about through reducing the power of the legislative veto players,
namely political parties, it is hardly surprising that few constitutions contain
provisions for such referendums. Overall, 11 constitutions23 allow for such
referendums, but many of these provisions, mostly in East European coun-
tries, have hardly been used. At the national level Switzerland uses such
popular initiatives most frequently. In several policy domains, for instance
environmental protection, this institution is credited with having triggered
signicant policy changes. A case in point is a constitutional amendmentin favor of protecting the Alpine region against automotive trafc through
the Alps, which Swiss voters adopted 20 February 1994, to the chagrin of
most elected ofcials. This constitutional amendment basically ruled out
the construction of any additional roads crossing the Alpine regions,
which might lead to increased trafc through or across the Alps. This new
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 489
22. Strictly speaking, the constitution in force required the parliament to agree to a referen-
dum, but the supreme court allowed the referendums to go forward which resulted in a consider-able overhaul of the constitution, giving the president increased powers, including the power to
launch a referendum on any question he chooses.
23. The following countries have such provisions: Belarus, Georgia, Latvia, Liechtenstein,
Lithuania, Moldova, Philippines, Slovakia, Slovenia, Switzerland and the Ukraine.
-
8/4/2019 Veto Players and Referendums Around the World
26/52
constitutional amendment originated in a signature petition proposing this
change to the constitution. Most observers were surprised by the outcome
since it endangered in part the delicate negotiations with the European
Union on bilateral agreements and transport.24
The remaining types of referendums, namely the required referendum and
the popular veto, leave the powers of the veto players in the representative
arena untouched, and thus these provisions decrease the potential for
policy change. Tables 1 and 2 suggest that this decrease in the potential for
policy change might reect the wishes of the framers of the constitution.
Often required referendums apply to constitutional amendments, changes
in borders, transfer of powers to international organizations etc. For such
policies, obviously, it is desirable to have a limited potential for change.
Similarly, popular vetoes often also apply to policy areas that are of centralimportance. Thus, for instance, the Danish constitution in article 20 requires
a majority of 83.3 percent of all members of parliament for bills involving
transfers of powers to supranational organizations. Thus, parliamentary
minorities of at least one-sixth can force a referendum vote, such as the
one leading to the negative outcome in the rst referendum on the Maastricht
Treaty (e.g. Ko nig and Hug, 2000). Again, provisions for such popular
vetoes decrease the potential for policy change, as the opt-outs negotiated
by the Danish Government after the Maastricht referendum clearly illus-
trate. A partial exception to this stability enhancing effect of popular vetoesappears with the practice in Italy labeled referendum abrogativo. Article 75
allows for referendums on enacted laws, but also allows for the elimination
of parts of laws. Since laws are often artful packages, allowing referendums
on elements of these packages also reduces, to some degree, the power of the
parliamentary veto players. A clear case in point is the Italian referendum of
19 April 1993 with which the Italian voters canceled a disposition in the
electoral rules employed for the election of the senate. More precisely, the
referendum largely eliminated proportional representation from the election
of the senate. As a consequence of this referendum, the government ofGiuliano Amato resigned and the resigning prime minister proclaimed the
end of the rst Republic (Newell and Bull, 1993: 607). In some sense it was
the culmination of the radical changes that the Italian political system had
undergone since the thorough judicial investigations of the partitocrazia
carried out under the label of mani pulite.
490 JOURNAL OF THEORETICAL POLITICS 14(4)
24. Bilateral agreements between the EU and Switzerland were negotiated in the aftermath ofSwiss citizens' rejection of the European Economic Area, a treaty concluded between countries
of the European Free Trade Association (EFTA) and the EU. Swiss voters accepted the bilateral
agreements on 5 May 2000, and at the time of writing EU member countries were in the process
of ratifying these agreements.
-
8/4/2019 Veto Players and Referendums Around the World
27/52
The investigations started by prosecutors from Milan divulged the massive
corruption of the Italian party system. Reformist forces attempted through
numerous referendums to force changes in the Italian system. Referendums
in Italy allow citizens to collect signatures to force a vote on most legislation
and parts of it, long after it has been adopted (Table 2). However, referen-
dum challenges face two important hurdles. First, they have to be accepted
by the constitutional court. Of the 13 referendums submitted to the court
in 1993, only 10 were allowed to proceed. Second, for a referendum outcome
to be decisive, the turnout has to exceed 50 percent. On 19 April 1993 this
threshold was easily passed with a turnout of 77 percent, and a yes vote of
82.73 percent. While the adopted referendum only struck a paragraph
from the electoral law for the senate, its consequences were considerable.
Parliament, which already had attempted unsuccessfully to fend off thereferendum by adopting its own changes, went into action and overhauled
the election laws in their entirety (D'Alimonte and Chiaramonte, 1993).
Elections became much less proportional and the plurality element came to
predominate. Only after these changes did alternation between two major
ideological blocks became possible.
Thus, only if the institutional rules allow parts of a bill to be struck down
in a referendum may the popular veto increase the potential for policy
change. If only entire laws can be struck down, policy stability never
decreases. Given that the potential for policy change differs under the varioustypes of institutions allowing for referendums, Gallagher's (1996: 237) mixed
assessment of the conservative bias of referendums is perfectly understand-
able. Depending on the precise institutional provisions referendums may
increase or decrease policy stability. Our model suggests clear links between
institutions and potential for policy change. Table 1, by listing all countries
with required referendums, indicates where we would expect referendums to
lead to increased policy stability (only in the policy areas where such referen-
dums are possible). The countries listed in Table 2, by allowing non-required
referendums to occur, often reduce the number of relevant veto players andthus mostly have reduced policy stability. But this reduced policy stability is
only to be expected if the agenda-setting in the referendum process is either in
the hands of non-veto players or a subset of the veto players.
Apart from predicting various degrees of stability as a function of the pre-
cise institutions allowing for referendums, our model also links these institu-
tions to policy consequences. Our analysis suggests that provisions for
referendums generally lead to policies more closely related to the preferences
of the voters. The effects of referendums differ, however, according to the
exact institutional provisions and which actors (veto players or non-veto players) the latter empower to set the referendum agenda. While the
information provided in Table 2 would allow for much more stringent
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 491
-
8/4/2019 Veto Players and Referendums Around the World
28/52
empirical tests of our theoretical predictions, existing empirical analyses of
the policy consequences of referendums largely omit the effects that different
congurations of veto players may have. Thus we can only relate our predic-
tions with respect to policy consequences on the basis of the classication of
referendums presented in Figure 4.
Despite this limitation the predictions based on our model strongly reso-
nate with most of the empirical evidence from systematic studies on the
policy effects of referendums. Given that provisions for popular initiatives
were present only in a handful of countries at the national level, few, if
any, studies have attempted or succeeded in demonstrating empirically the
effect of such referendums on policy outcomes in cross-national analyses.
In studies focusing on the effect of institutional provisions at the subnational
level, both in the United States and Switzerland, the results suggest that pro-visions for popular initiatives and, to a lesser extent, provisions for popular
referendums lead to policies closer to the voters' ideal-point. Gerber (1996,
1999) shows for two policies, namely laws on parental consent for teenage
abortion and the death penalty, that in states with the popular initiative
the policies adopted correspond more closely to the voters' preferences
than in states without the initiative. Similar effects appear in Gerber and
Hug (2001) for a series of policies aiming at protecting minorities. Most
closely related to the theoretical insights discussed in this paper are Hug's
(1999) empirical results. Replicating and extending Gerber's (1999) work,he demonstrates that the voters' preferences are the most closely reected
in the presence or absence of the death penalty in states that allow for
popular initiatives. The relationship between voter preferences and policy
outcome in states with the popular veto is weaker, and still weaker is the
relationship in states with no non-required referendums. Only in the work
by Lascher et al. (1996) and Camobreco (1998) do authors report evidence
counter to our theoretical results. Matsusaka (2001), however, convincingly
challenges the empirical strategies that these authors employ.
Only partly related to this work attempting to demonstrate the effect ofvoter preferences on policy outcomes is the literature dealing with the
effect of provisions for referendums on economic outcomes. Several studies
show that initiative states (and cantons) have lower tax burdens, lower
decits, etc.25 Since voter preferences do not enter directly into these empiri-
cal models, they only indirectly test propositions related to the theoretical
conclusions of this paper.
492 JOURNAL OF THEORETICAL POLITICS 14(4)
25. Kirchga ssner et al. (1999) review in detail these studies, and Matsusaka (1995, 2000) pro-
vides discussion of the literature on the American states.
-
8/4/2019 Veto Players and Referendums Around the World
29/52
Conclusion
The purpose of this paper was to organize our understanding of referendums
and the institutions that regulate them in a theoretically consistent andempirically accurate way. We saw that the overall assessments of referen-
dums have been controversial: for some authors referendums are the expres-
sion of popular will, for others the connection between popular will and
referendum outcome cannot be established. We also saw that classications
of referendums do not use the same criteria, and that some of these criteria
are based on strategic considerations (like who takes the initiative) while
others are based on obscure principles (pro- or anti-hegemonic, controlled
or uncontrolled). Finally, even the theoretical approaches come to different
conclusions depending on the dimensionality of the underlying space: single-dimensional approaches can calculate the results of a referendum as a func-
tion of the preferences of different actors; multi-dimensional approaches
stress cycles, paradoxes and inadequacies in outcomes.
In order to deal with all these variations in assessments and approaches, we
introduced a multi-dimensional model which enabled us to understand and
explain all these discrepancies. Our model was based on the theory of veto
players, which identies the individual or collective actors that are required
to agree for a change of the status quo. On the basis of veto player theory,
we concluded that referendums introduce one additional veto player, themedian voter of the population (or a very close approximation of it, the
center of the yolk of the population). In addition, a circle very well approx-
imates the outer bound on outcomes that are preferred over the status quo by
this new veto player, as if the millions of voters participating in a referendum
were a simple individual.
On the basis of our model, the introduction of the mere possibility of a
referendum shifts the outcomes of legislative politics closer to the population
median. The reason is that legislative outcomes that would replace the status
quo by a point that the median voter dislikes more than the status quo canimmediately be defeated by a referendum, so successful policy choices have
to be located inside the set of preferences of the median voter. Empirical
studies corroborate this expectation. Viewing referendums as the introduc-
tion of one additional veto player in a political system leads to the prediction
that the stability of outcomes increases.
The other major consequence of our model is the identication of the insti-
tutions that shape not only referendum outcomes, but also the outcomes of
the legislative game when referendums are possible. On the basis of our
analysis the institutions regulating agenda-setting in referendums signi-cantly affect the possible outcomes. Agenda-setting in referendums is divided
in two parts. The rst is who triggers the referendum (required referendum,
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 493
-
8/4/2019 Veto Players and Referendums Around the World
30/52
existing veto player or non-veto player). The second is who asks the question
(veto player, non-veto player). On the basis of the answers in these two ques-
tions referendums are divided into four different categories: required referen-
dums (where the question is asked by veto players), veto player referendums
(veto players both trigger and ask the question of the referendum), popular
vetoes (non-veto players trigger, but veto players ask the question), and
popular initiatives (non-veto players both trigger and ask the question).
The more agenda-setting is divided, the more likely that competition
among the different actors involved may lead the referendum results to
approximate the preferences of the median voter. For this reason, popular
initiative institutions should be studied on the basis of by whom and under
what conditions issues can be placed on the ballot. We believe that the differ-
ences in assessments of referendums existing in the literature (as discussed inthe introduction of this article) can be attributed to the differences in agenda-
setting processes. People who dislike referendums pay attention to the ques-
tion of who controls the agenda of a referendum and why this actor triggered
it; people who like referendums assume a competitive popular initiative pro-
cess.
At the empirical level, existing systematic studies largely support the
theoretical implications we derive. Institutions allowing for referendums to
be triggered by veto players or non-veto players lead to policies that reect
more closely the voters' preferences. The strongest additional effect of voterpreferences on policy outcomes appears under institutional provisions allow-
ing for popular initiatives. However, as Hug (1999) shows, this stronger
effect diminishes quite considerably as the signature requirement increases.
But decreasing signature requirements also corresponds with an increasing
number of referendums. Thus, constitutional framers should balance well
between different institutional provisions for referendums and the costs
they wish to impose on possible users of these institutions.
While existing empirical research lends some credence to our theoretical
model, future research should carry out more stringent tests. More precisely,the information provided in Table 2 on the veto players in control of the
referendum agenda in non-required referendums allows for much more
precise predictions of the policy consequences, even on multi-dimensional
issues. In particular, if provisions for referendums cancel out other veto
players our model would predict policy outcomes much more heavily inu-
enced by the veto player in control of the referendum agenda. Such stringent
tests would give constitutional framers much more material on which to
base their decision whether to introduce provisions for referendums in new
constitutions.
494 JOURNAL OF THEORETICAL POLITICS 14(4)
-
8/4/2019 Veto Players and Referendums Around the World
31/52
APPENDIX
HUG & TSEBELIS: VETO-PLAYERS & REFERENDUMS AROUND THE WORLD 495
Table 1. Required Referendums Around the World
Country Constitutional Provision for Required Referendum Constitution
Used
(revised)
Antigua-Barbuda Art. 47, constitutional amendments, two-thirds of all the
members of the House must approve amendment
1981
Australia Art. 128, constitutional amendments, law must be passed
by an absolute majority of each House of the Parliament
or by one House of Parliament and submitted by the
governor-general
1901
Austria Art. 44, total revision of constitution after adoption inLower House and absence of objection in Upper House
1929
Bahamas Art. 54, constitutional amendments, after two-thirds
(three-quarters for some Articles) majorities in both
Houses adopted amendment
1973
Bangladesh Art. 142, constitutional amendments, a bill, passed by
two-thirds majorities, which provides for the amendment
of the Preamble or any provisions of Art. 8, 48, 94[0r],
56, 95 or 142
1996
Botswana Art. 89, constitutional amendments, after passage by
two-thirds majority in the Assembly
1966
Colombia Art. 377, constitutional amendments to Ch. 1 of Title II,
after passage in Parliament
1991 (2001)
Congo Art. 172, international treaties for cession, exchange or
addition of territory proposed by the president must be
adopted by referendum
1979 (1992)
Croatia Art. 135, for the association of the Republic of Croatia
in alliances with other states, after passage in Parliament
by a two-thirds majority vote
1990
Cuba Art. 137, constitutional amendments, after vote by
two-thirds in National Assembly, provided theconstitution is completely revised, or the amendment
refers to the powers of the National Assembly or to the
rights and duties established in the constitution
1976
Denmark Art. 88, constitutional amendments, after passage by
two separately elected Parliaments
Art. 28, changes in the voting age, after passage by
Parliament
1953
Dominica Art. 42, constitutional amendments, after approval by a
three-quarters majority in Parliament
1978
Egypt Art. 189, constitutional amendments, after approval bytwo-thirds of the members of the Assembly
1980
continued on next page
-
8/4/2019 Veto Players and Referendums Around the World
32/52
496 JOURNAL OF THEORETICAL POLITICS 14(4)
El Salvador Art. 89, constitutional amendments on the creation of a
Central American republic after passage by Parliament
1983
Gabon Art. 113, international treaties, if they involve territorial
changes
1975 (1991)
Gambia Art. 226, constitutional amendments, after passage by
three-quarters majority in Assembly
1970 (1996)
Grenada Art. 39, constitutional amendments, after passage by
two-thirds of the members of the House, amendments to
some sections of the constitution have to be adopted in
a referendum
1974
Guatemala Art. 280, 173, constitutional amendments, after a
two-thirds vote to amend constitution in the National
Assembly
1985
Guyana Art. 164, constitutional amendments, after majority of
Parliament has approved amendment to certain parts of
the constitution
1980
Iceland Art. 79, constitutional amendments, after passage by
Parliament of an amendment to the status of the
Church under Art. 62
1944
Ireland Art. 46, constitutional amendments, initiated in the
Lower House and after passage in both Houses
1937
Jamaica Art. 49, constitutional amendments, after passage by
two-thirds majorities in both Houses
1962
Japan Art. 96, constitutional amendments, initiated in the
Diet and after passage by two-thirds majorities in both
Houses
1964
Kiribati Art. 69, constitutional amendments, after passage by a
two-thirds majority in Parliament
1979
Korea, South Art. 130, constitutional amendments, after passage by a
two-thirds majority in the National Assembly
1987
Latvia Art. 76, 77, amendments to Art. 1, 2, 3, 4, 6, or 77 of the
Constitution, after passage by two-thirds majority in
Parliament
1922 (1998)
Liberia Art. 91, constitutional amendments, after passage by
two-thirds majorities in both Houses
1984
Lithuania Art. 148, amendments to Art. 1, Chs 1 and 14 of the
constitution, after passage by two-thirds majority in
Parliament
1992
Macedonia Art. 74, 120 changes of borders or association with or
dissociation from a union, after passage by Parliamen