1

Decision-making power in the European Unions

Council of Ministers: New Empirical Findings and
Suggestions for a Research Agenda
Paper prepared for the 2005 MPSA Conference
7th-10th April
Chicago
Sara Hagemann
s.hagemann@lse.ac.uk
Department of Government
London School of Economics and Political Science
Abstract
This paper sees coalition formation in the European Unions Council of Ministers as a
bargaining game between strategic, self-interested actors who conform into policy-
connected coalitions based on the distribution of voting power. By testing this
hypothesis together with a set of variables which in the literature are often assumed to
affect the member states voting behaviour, it is found that only a countrys left/right
ideology, its voting power and whether or not it is holding the Presidency are of
significance for its propensity to oppose the majority at voting stages prior to the final
adoption. Only at the last voting stage does a countrys role as a contributor or
beneficiary to the EU budget also matter. These are controversial and important new
findings and appear when analysing a data set covering all legislation passed from
1999 to May 2004 and consisting of votes from all stages of the legislative process.
The results indicate that the member states voting behaviour change across the
various stages of the legislative process and that small and big member states use
different strategies to influence policy outcomes.
2
Introduction
Research on legislative politics in the European Union has for several decades
focused on the inter-institutional bargaining between the European Parliament and the
Council of Ministers. Only recently have EU scholars turned their attention to the
internal dynamics of the two institutions and provided insights into the policy
processes within each body. Yet, whereas research on the EP is now an established
sub-field of the study of EU legislative politics, the study of Council decision-making
still suffers from a lack of rigorous empirical analysis as well as a common theoretical
framework.
The aim of this paper is to further our empirical knowledge of Council decision-
making by investigating a data set consisting of all votes taken in the Council from
1999 to May 2004 and covering all stages of the legislative process. The main
concern is to update our knowledge on which factors have an influence on the
member states voting behaviour and also to look into whether their voting behaviour
changes across the different stages of the legislative process. Furthermore, based on
recent theoretical insights from the literature, I present coalition formation in the
Council as a process where ideology, national political issues and other such factors
are thought to influence the direction of decision-making. However, as it is the
respective countries power that in the end makes the adoption or rejection of a policy
proposal possible, I suggest that this is what ultimately determines the coalition
building and voting outcomes. To be specific, I present legislative politics in the
Council as a bargaining game between strategic, self-interested actors who conform
into policy-connected coalitions based on the distribution of voting power
1
. This
hypothesis is tested on the data together with a number of other assumptions often
found in the literature.
Decision-making in the Council
The data used for the analysis in this paper included a few surprises, suggesting that
the current descriptions of Council decision-making may need to be updated. The
1
Will be defined below.
3
following is therefore a short outline of existing knowledge, which is then elaborated
upon in the light of the findings from the 1999-2004 data.
The EU legislative process is formally started when the Commission issues a policy
proposal to the EP and the Council. The procedure for adopting the proposal depends
on the legal basis of the initiative, which the Commission determines in the
preparatory stages. When a policy proposal has been initiated and presented to the
Council it is usually first discussed in specialised working groups where officials from
the member states and the Commission meet. Gradually, proposals advance through
the preparatory bodies closer to the Council. The most senior of the preparatory
committees is the Committee of Permanent Representatives (COREPER) from where
proposals are sent to the Council as either A or B points. From previous
descriptions of Council decision-making (e.g. Hayes-Renshaw and Wallace 1997;
Sherrington 2000) we have been told, that at this stage A points are normally agreed
upon already and therefore accepted without much discussion in the Council. The
more controversial agenda items are categorised as B points
2
.
When voting takes place, different rules apply depending on the policy area:
unanimity is applied to certain matters affecting the members fundamental
sovereignty, and a weighted qualified majority system (QMV) to others. The key
features of the QMV system is that all members have a seat but that their respective
number of votes vary, reflecting the differences in population shares (Leech
2002:439). Appendix 1 shows the distributions and thresholds for the QMV systems
throughout the EUs history.
However, the Council does not always vote in the formal sense of raising hands
(Westlake 1995:87). A proposal can be adopted with a single sentence from the
Chairperson when she knows that there is a sufficient majority or unanimity in the
Council. If the Council is not unanimous, the Chairperson still knows the member
states positions from the preceding talks in the preparatory bodies (Mattila 2004:30).
If she decides that enough member states have come to an agreement to meet the
threshold, the proposal is accepted and those opposing or abstaining can record their
views officially. Yet, one important difference between the unanimity and QMV
systems must be pointed out. When the decision rule is unanimity, abstentions are not
2
Van Schendelen (1998) finds that more than half of all decisions made by the Council are categorised
as A points.
4
counted as no votes. This means that decisions can be made with few countries
actually voting for the proposal, if none of the countries actively opposes it. The
opposite is true for QMV, where the high threshold makes abstentions have the same
effect as no votes in practise.
Depending on the decision-making procedure the Council may be presented with a
proposal and its various amendments several times during the process that leads to its
final adoption or rejection. For example, in the Co-decision procedure, the Council
may adopt a common position before the proposal goes to the European Parliament
for a next reading after which it may return to the Council once again. The current
literature has not explained in detail what the implications of this form of sequential
bargaining may be for the internal dynamics in the Council, but I find it reasonable to
assume that the member states voting behaviour will change in accordance with the
exchange of information, vote trading and strategies that prevail throughout the
different stages of the negotiations. For example, it seems plausible that at the last
stage, where the member states decide whether to adopt or reject a proposal
altogether, only certain member states are willing to oppose the majority (Mattila
2004; Mattila and Lane 2001). The cost of opposing at this stage is considerably
higher and also the chance of media coverage may influence the governments, thus
making voting behaviour at this point different from the amendment voting at earlier
readings.
What we didnt know about decision-making in the Council
From the data it is apparent that particularly three issues need to be re-evaluated in the
current literatures description of Council decision-making. First, there is the
distinction between A and B agenda points. Although it is true that the intention
behind the A and B points has traditionally been to ease the workload at the
ministerial level and lower the need for bargaining and formal voting at the last stage
of the legislative process, the data shows a very rare use of the B point category
which has traditionally been for controversial agenda items only (please refer to Table
1). Particularly the last few years have had an extremely low amount of B points
with only 6.7% and 1.7% in 2003 and 2004, respectively. The lack of B points
could mean one out of two alternatives: 1) either it lends support to the consensus-
5
oriented branch of the literature
3
and indicates that the member states are actually able
to settle agreements without any disputes at the ministerial level; or 2) the difference
between A and B items is diminishing. I have addressed this issue elsewhere
(Hagemann 2005) and found that the low level of B items is most likely due to the
fact that open disputes are increasingly accepted under A points as well, essentially
making the B point category superfluous. Also, the fact that not all B items are
actually a basis for dispute supports the suspicion that the difference between A and
B points may be declining. In 2000 and 2001 there was only disagreement over
36% and 33% of the B points, respectively.
Table 1
Year
Number
of acts
adopted
B
points
Disagreement*/B
points
1999 161
10
(6.2%)
70%
2000 169
25
(14.8%)
36%
2001 160
15
(9.3%)
33%
2002 163
17
(10.4%)
88%
2003 163
11
(6.7%)
82%
2004 115
2
(1.7%)
100%
* By disagreement it is meant that voting took place and
more than 2 countries opposed or abstained from voting
Only legislation adopted before May 1
st
of this year is
included due to the enlargement
Secondly, a very important but largely unrecognised issue is how the Amsterdam
Treaty may have changed decision-making within the Council. After the reform of the
Co-decision procedure in the Amsterdam Treaty, which came into effect in May 1999,
more legislation falls within the Co-decision procedure and can now be adopted
already in the first reading in the Council. In other words, it is currently possible to
see Co-decision legislation adopted at either the first reading in the Council, the
second reading in the EP, the second reading in the Council or in the Conciliation
3
For an overview see e.g. Heisenberg 2005.
6
Committee
4
. Table 2 summarises the findings regarding the voting rules and the
amount of QMV legislation adopted at the Councils first reading.
Table 2
Year
Number of
acts
adopted
B points
Council Voting
Rule, Unanimity
Council
Voting Rule,
QMV
QMV legislation
adopted in
Council 1st
reading
1999 161
10
(6.2%)
85
(52.8%)
76
(47.2%)
41
(53.9%)
2000 169
25
(14.8%)
80
(47.3%)
89
(52.7%)
52
(58.4%)
2001 160
15
(9.3%)
58
(36.3%)
102
(63.8%)
32
(31.3%)
2002 163
17
(10.4%)
66
(40.5%)
97
(59.5%)
35
(36.1%)
2003 163
11
(6.7%)
103
(63.2%)
60
(36.8%)
35
(58.3%)
2004 115
2
(1.7%)
87
(75.7%)
28
(24.3%)
19
(67.8%)
(%) are row percentages.
Only legislation adopted before May 1
st
of this year is included due to the enlargement
The possibility for adopting legislation already at the Councils first reading means
that the overall legislative process can no longer be described as a sequential
bargaining course, but may rather be dominated by simultaneous moves in the two
institutions in an effort to speed up the process. Certain changes have therefore also
occurred concerning both the internal organisation of the Council as well as in the
relationship with the EP.
With regard to the internal procedures in the Council, especially two issues are of
importance: the use of formal procedures for reaching agreements and the role of the
Chair. When comparing the data from this paper with earlier accounts of the use of
formal procedures in Council decision-making (for an overview see e.g. Heisenberg
2005) it becomes clear that, most likely due to time pressure, formal rules are
increasingly used for reaching agreements, effectively making decision-making more
efficient. A very interesting finding related hereto is that there has especially been a
significant increase in the member states use of formal statements immediately
following the adoption of a policy proposal. Formal statements have in the literature
4
For an explanation of the EUs inter-institutional legislative process see e.g.Hix 2005.
7
traditionally been described as only being used for when a country has opposed or
abstained on a proposal and wishes to make its reasons for doing so public. Yet, the
data used in this paper indicates that these statements are not only used as a
supplement to the votes: They often show strong disagreements with the majority
even in cases where the opposing country does not show its dissatisfaction through
the votes. It therefore requires some explanation why member states may have
reasons not to oppose the majority by voting, but instead choose to make their
opposition clear in the formal statements. I will return to this matter in the analysis
below but here just note that, contrary to conventional wisdom in the field, since 2001
each year has experienced that more than 65% of legislation could not be agreed upon
by consensus when including both votes and formal statements (Hagemann 2005).
Beside of the increased use of formal statements and votes, the changes in the macro-
institutional rules introduced by the Amsterdam Treaty can be assumed to lead to
other changes in individual actors respective influence over policy outcomes (Farrell
and Heritier 2004:5). For example, there may be a need for adjustments of such
positions as the chairmanship. A skillful Chairperson in the Council (i.e. a
representative for the country currently holding the Presidency) could, for instance,
due to hard deadlines be left with a great potential for exploiting information
asymmetries, a skewed distribution of strong and weak preferences over a policy
proposal etc. This is due to the fact that, although a potential agency-drift is apparent,
other member states may prefer to rely on the Chairperson with regard to not only the
handling of the internal negotiations, but also with regard to the coordination and
bargaining with the EP
5
. Since the inter-institutional bargaining is no longer purely
characterised by sequential moves and there is a greater pressure from a shorter policy
process as well as hard deadlines for negotiations in each of the institutions, a higher
level of coordination across the two institutions is necessary. Here, the Chairperson
can take advantage of her national party organisation and strengthen the Councils
position vis--vis the EP by gathering information on the EP party groups
preferences through MEPs representing the same national party
6
. This may prove to
be a useful coordination mechanism indeed, and if one was to speculate one step
5
For simplicity, the Commission is disregarded for now, but could of course be taken into account as
well.
6
One example where the role of the Chair and the EPs rapporteur was of great importance was in the
2003 amendment of the Directive 94/62/EC on packaging and packaging waste. Please refer to
http://europa.eu.int/abc/doc/off/bull/en/200312/p104044.htm
8
further, the fact that both the EP and Council are made up of political parties (Hix
2002), makes it convincing that representatives in the Council seek to harmonise the
policy preferences of MEPs from governing parties with the preferences of a
sufficient majority in the Council
7
. Needless to say, further research on this topic is
required, yet, it should by now be clear that we cannot by default assume that the
Council is a unitary actor vis--vis the EP, as is so often supposed in the literature.
Hypotheses
Taking into consideration the above observations, decision-making in the Council is
best presented as a repeated bargaining game between self-interested, rational actors
(the member states) possessing perfect information about other players policy
preferences. The member states engage in negotiations over policy proposals initiated
by the Commission in order to pursue own policy goals. The incentive for the actors
to participate is a combination of the opportunity to influence politics beyond own
territorial and/or political boundaries and the wish to maintain control of domestic
politics and interferences by foreign EU governments. Coalitions can based on this
be assumed to form between actors with similar policy preferences (Axelrod 1970)
and in compliance with the formal rules for adopting legislation.
The importance of the formal rules for coalition formation has been a matter of great
dispute (e.g. Heisenberg). However, especially when considering the increase in the
use of voting and formal statements (Hagemann 2005) it is still reasonable to assume
that all policy outcomes are indeed agreed upon with the distribution of voting
weights in mind (see also Golub 1999). Yet, as voting weights do not accurately
reflect a countrys control of decision-making (e.g. Penrose 1946)
8
, I will here rely on
the insights from the voting power index literature and hypothesise that it is the
7
Whether or not sufficient majority means a supermajority reached by consensus or only the
required qualified majority as reached by voting is irrelevant for the argument here.
8
A classic example which illustrates the fact that voting weights are not comparable with voting power
is when Luxembourg, with its one vote in the first phase of the Councils history, turned out to have
absolutely no voting power over any decision taken by QMV despite the disproportional representation
of its population. Since the threshold required to reach a decision was 12 votes during this period, it
was mathematically impossible for Luxembourgs one vote to be decisive: no matter how the five other
countries voted, their combined total would never be equal to 11 (Brams and Affuso 1985; Leech
2003). Similar situations are still possible and make it necessary to turn to power indices, which
calculate each member states frequency of being pivotal in voting outcomes across all logically
possible combinations of votes.
9
respective countries voting power which ultimately determines coalition building and
voting outcomes. Voting power indices measure each countrys relative influence on
decision-making in terms of a priori power established by the formal voting rules and,
contrary to the distribution of voting weights which are given in absolute figures,
make it possible to directly compare those values across the member states.
Voting power can be thought to affect voting behaviour in one of two possible ways
(Mattila 2004:33). First, one could hypothesise that countries with more voting power
dominate voting situations, that they are able to manipulate the voting outcomes
towards their own preferred policy point and that they are more likely to be on the
winning side of any vote in the Council. As officials are mainly concerned with
getting proposals adopted and therefore pay a special attention to the preferences of
the larger member states
9
, those countries are also likely to have more influence in the
preparatory stages of any decision, and make this argument seem reasonable.
On the other hand, recent research has shown that large countries vote against the
majority more often than smaller countries at the last stage of the legislative
procedure (Mattila and Lane 2001; Mattila 2004). This may reflect strategic
differences in the approach to decision-making. Smaller countries may acknowledge
their limited resources and abilities to influence every decision made by the EU. Thus,
they restrict their attention to issues that they consider especially important, or they
may seek to influence legislation earlier in the process than the last voting situation.
Also, larger countries may be more likely to feel the need to voice their opinions more
forcefully if they disagree; their electorates could find it very difficult to have their
national interests overruled by EU decisions. In this situation, the government can
alleviate its difficult position by voicing its dissent publicly although its opposition
will not affect the final outcome (Mattila 2004:33).
In the empirical analysis I measure a set of variables affect on the likelihood of a
country opposing the majority. Two data sets are used for this purpose, one consisting
of the official votes from the last possible vote of 931 policy proposals and another set
consisting of 349 votes on a sample of the same policy proposals, but only including
voting situations prior to the final stage of the decision-making process. The total
9
Therefore, they may act according to the logic of backward induction.
10
number of observations for each data set is 13.965 and 5.235, respectively. If the first
example above about voting powers affect on member states behaviour is true, none
of the large countries should be found to be likely to oppose the majority. If the latter
example is true, we can expect that only the large member states should oppose when
we focus on the last possible vote. Yet, when voting situations prior to the last votes
are analysed, small and big member states should be equally likely to be in
opposition. Hence, Hypothesis 1 and 2:
H1: A country with more voting power will oppose the majority more
frequently than a country with less voting power at the last
voting stage.
H2: Voting behaviour in the Council changes across the different stages of the
legislative process
Unlike research on voting in the European Parliament (e.g. Hix 1999) or on the
general inter-institutional decision-making in the EU (e.g. Tsebelis and Kreppel
1998), no unifying political space has been identified as the framework within which
the member states in the Council act. Though, adopted from the general EU studies, it
is broadly assumed that also internally in the Council at least two dimensions exist: a
left-right dimension as well as a dimension capturing how far European integration
should progress at the expense of the nation state (Hooghe and Marks 1999; Hix
1999; Marks and Steenbergen 2004). I assume that present EU politics are
predominantly fought along the left/right dimension due to the already very elaborate
cooperation of the member states and intermingling of national and EU political
matters. Also, as already mentioned, one should not disregard the fact that, besides of
being national representatives, the actors in the Council are also political parties with
policy goals along the traditional left/right axis. I therefore expect to find that the
rational governments form connected-winning coalitions in the way that centre-left
and centre-right governments conform into separate groups of potential alliances
including governments within these groups according to the necessary majority
requirement. Consequently, as the majority of governments in the period studied here
consist of centre-left or left wing governments, I expect that right-wing governments
will be found to oppose the majority more often than their left-wing counterparts.
Thus, Hypothesis 3:
11
H3: A right-wing member state is more likely to oppose the majority than a
left-wing member state.
An effect of the left-right ideology variable can be assumed to be detected both when
analysing the last possible vote and when analysing votes from earlier stages of the
legislative process.
Variables
For the purpose of the empirical test, the above three hypotheses are operationalised
in the following way: the first hypothesis considering the member states voting power
is calculated on the basis of the normalised Banzhaf Index (Banzhaf 1965) by using
the POWERSLAVE (2002) programme. Please refer to Appendix 2 for an
explanation of the normalised Banzhaf Index. The second hypothesis regarding the
changes in voting behaviour across the different stages of the legislative process can
be concluded upon by comparing the results from the two data sets. The third variable
locating the incumbent governments positions on the left-right ideology dimension is
measured with an index variable that ranges from 0 to 1. The variable consists of each
governments policy position as found in Gabel and Huber (2000) and has been
divided into 6 months units because of the Presidency-variable mentioned below.
The government that is furthest to the right has the value of 1 and the government that
is furthest to the left has the value of 0.
In the literature, other factors than the voting weights and voting power are often
highlighted as having an important effect on voting behaviour and coalition formation
in the Council. I have included these variables in the analysis in order to explore the
broader picture of what truly makes up Council voting behaviour
10
. The variables are
the following
11
:
1. The first additional variable is the governments position on the above mentioned
more/less integration scale calculated similarly to the left/right variable and as
10
For detailed literature reviews, which, among other things, examine each of these variables, please
see Zimmer et al (2004) and Selck (2004).
11
It should be noted that, although the empirical findings are quite different from the ones presented in
this paper, some of the variables are also to be found in Mattila (2004) (See
http://www.valt.helsinki.fi/staff/mmattila/contest/) and and Heisenberg (2005).
12
found in Marks and Steenbergen (2004). A high score indicates an attitude
favourable to European integration and a low score indicates Euro-scepticism. I
predict that the more/less integration variable does not have an effect when
analysing votes from earlier readings nor when confining the analysis to the last
possible votes. I find it convincing that the EU system has indeed developed into a
political system of its own (Hix 1999:5), largely substituting the traditional
intergovernmentalist view of an integration/status quo divide with a left/right
cleavage as known from national political systems.
2. It is quite possible that an interaction effect exists between the two variables
capturing the policy dimensions described above. For example, Euro-sceptical
left-wing governments may behave differently to Euro-sceptical right-wing
governments (Mattila 2004:41). Therefore, a variable is included to see if there is
a significant impact of this combination on governments voting behaviour (see
also Hooghe and Marks 1999). Similar as to the left/right variable, I predict that
the interaction variable left/right ideology x support for integration affects
whether or not a government will oppose the majority at all stages of the decision-
making process.
3. Whether or not an EU member falls into the category of receiver or
contributor to the EU budget is also often argued to affect the likelihood of
being either in opposition or in favour of a proposal. Certain theorists even argue
that EU politics in general is about the wealthier member states pay-offs of
poorer nations by means of subsidies (e.g. Carruba 1997; Hosli 1996). I
acknowledge this phenomenon in EU politics, yet perhaps not to a similar extent
as some of the previous studies of Council decision-making, and do not expect
this phenomenon to be apparent in the analysis, at least not at the last voting stage.
The pay-offs and vote trading will basically have taken place before the votes on
redistributive issues are voted upon. Therefore, I predict that whether or not a
government is a contributor or beneficiary to the EU budget does not have an
effect on the likelihood of opposing the majority at the last voting stage nor in
voting situations at earlier readings.
4. A variable measuring whether there is an effect of holding the Presidency is also
included. The literature concerning this matter is mostly descriptive (Tallberg
2003) and it has in many cases proven difficult to establish exactly what the
consequences and opportunities are for a country holding this position, as was also
13
reflected in the above discussion of the role of the chairperson. Here, we will at
least be able to establish what the consequences are for the likelihood of opposing
the majority. The variable is included as a dummy variable and I predict that
whether or not a country is holding the Presidency affects its likelihood of
opposing the majority when analysing voting situations prior to the final vote as
well as when only taking into account the last possible vote. I find it likely that
countries in general wish to send a signal of consensus when they are holding the
Presidency and therefore deliberately do not oppose the majority. Also, although it
has not yet been definitively established whether or not the Presidency directly
allocates a country with gate-keeping or agenda-setting powers or demands the
role of an honest broker, it seems likely that the Presidency position in general
disencourages any given member state from going against the Councils common
will.
5. A variable measuring the potential effect of the 2000 enlargement negotiations is
included as a control variable. This variable is meant to capture the unusual low
level of disagreements and QMV votes in 2000
12
, which are most likely connected
to the important Copenhagen agreement leading to the 2004 enlargement. Failing
to control for this could produce strongly misleading results.
6. As a last control variable I have included the number of decisions taken by the
Council. This is calculated as the natural logarithm of the total number of votes
taken in each half-year period and is used to control for the possibility that
whether or not a country is likely to be in opposition is affected by the total
number of decisions made (Mattila 2004:41). I predict that the number of
decisions taken by the Council does not have an effect on whether or not a country
will oppose the majority. The preparatory bodies should prevent any such effect.
In Table 3 I have summarised my predictions regarding which variables affect
whether or not a country will oppose the majority in the Council when analysing the
last possible vote and votes from previous voting situations, respectively.
12
Especially the last half year of 2000 shows a very low level of formal votes and statements.
14
Table 3: Predictions of variables effect on
frequency of opposing the majority
+ indicates a positive effect is expected.
indicates a negative effect is expected.
Data
As already mentioned, two data sets are used in the empirical test. Both of the data
sets cover the instances where the Council acts as a legislator. The first data set
includes all legislation adopted between 1999 and 2004 and has a total of 931 pieces
of legislation. It consists of the last possible votes on each proposal, that is, the
decisions where it is determined whether a final proposal is adopted or rejected. The
second data set consists of a sample of the votes from the same data set as above, but
only includes the votes from earlier readings. Yet, the number of acts prior to the final
vote includes only 349 pieces of legislation, and is therefore considerably smaller
than the first data set. This is partly due to the fact that not all of the legislation falls
within the Co-decision procedure and therefore does not include several readings. It is
also partly due to the fact that, out of the Co-decision legislation, 214 acts were
already adopted after the first reading and therefore had no previous voting situations.
However, the total number of observations in this data set still amounts to 5235 (15 x
349) and is indeed suitable for an analysis of voting behaviour across the different
stages of the legislative process.
Variable Last vote only Votes from
earlier
stages
left/right ideology
Governments support
for integration
Left/right ideology x
support for integration
Presidency
Contributor/Beneficiary
Voting power
+
No effect
+
-
No effect
+
+
No effect
+
-
No effect
No effect
15
As already explained above, the use of formal statements immediately following the
adoption of an act may actually be a way to voice a countrys opposition against a
proposal, while there may be reasons for not doing so by voting. I have therefore also
coded information on those countries who have stated they wished to oppose or have
given strong negative comments on a policy proposal such that a comparison can be
made between these official statements and the member states use of voting (for a
discussion of this see also Carruba et al. 2004).
2.4 Findings
Table 4 and 5 below show the frequency of oppositions, abstentions and use of formal
statements per country. It is apparent that the largest member states abstain or oppose
the majority more often than the smallest member states at all stages of the legislative
process (column 3 in both tables). However, compared to their share of opposing the
majority through voting, it seems as if the smaller member states to a greater extent
show their dissatisfaction through the use of formal statements than the big member
states do. At the last voting stage, the four smallest members oppose the majority in
the formal statements with an average of 15 times, whereas they only chose to do so
6.5 times by voting. The four largest member states, on the other hand, oppose or
abstain the majority through voting by an average of 30.5 times, whereas they on
average make formal statements 31.75 times. Therefore, when looking at the total of
abstentions, oppositions and formal statements instead of only the number of
abstentions and oppositions, the gap between large and small countries is not as
profound, although indeed still considerable.
At the stages prior to the final adoption (Table 5) the perhaps most remarkable finding
is that although the number of legislation has made a considerable drop from 931 to
349, the frequency of opposing, abstaining or making formal statements have not
decreased to a similar extent. The three largest countries still chose to oppose or
abstain from voting in 17, 11 and 24 cases, respectively. Also, in this table we see a
similar tendency as in Table 4, where the larger member states have a greater share of
the number of times they are in opposition or abstain, yet again, the use of formal
statements does not have the same dramatic decrease: both small- and medium sized
countries often make their opposition explicit through the use of formal statements.
16
Table 4: Last stage
Country In favour
Oppose
or
abstain
Formal
Statements
Opp.,
abst. and
statements
in total
Germany 864 41 16 57
France 892 32 7 39
UK 846 27 18 45
Italy 873 22 26 48
Spain 884 22 19 41
Netherlands 890 19 17 36
Greece 904 15 12 27
Belgium 891 9 14 23
Portugal 892 11 38 49
Sweden 885 9 27 36
Austria 891 5 9 14
Denmark 884 11 32 43
Finland 912 4 15 19
Ireland 907 8 7 15
Luxembourg 912 3 6 9
Table 5: Stages prior to final adoption.
Country In favour
Oppose
or
abstain
Formal
statements
Opp.,
abst. and
statements
in total
Germany 300 17 19 36
France 318 11 11 22
UK 295 21 33 54
Italy 307 22 14 36
Spain 308 14 17 31
Netherlands 330 8 11 19
Greece 327 11 18 29
Belgium 337 4 8 12
Portugal 332 9 19 28
Sweden 337 4 26 30
Austria 342 4 3 7
Denmark 330 11 31 42
Finland 337 2 14 16
Ireland 338 6 5 11
Luxembourg 340 2 7 9
In sum, the two tables show that 1) when taking into account the number of legislation
analysed, a higher percentage of disagreement is apparent at readings prior to the final
adoption of a policy proposal than at the last voting stage; and 2) although the largest
17
member states has the biggest share of opposition, abstention and formal statements,
there seems to be a difference in the strategies for how small and large member states
show their discontent. Small member states do so through formal statements more
than through voting, whereas the big member states use both voting and formal
statements.
Table 6 and 7 take us a step further and present the regression results of which of the
variables have an effect on the decision to oppose the majority. The dependent
variable is therefore the frequency of opposing the majority, that is, the sum of
abstentions, formal statements and votes cast against the majority in the Council
regardless of whether the latter is in favour of or suggests to reject the individual
policy proposals
13
. The results are generated by using a random-effects Poisson
regression model which is most suitable for this kind of event count variables
14
.
Table 6: Poisson regression last stage votes only
Random-effect Poisson regression Number of obs = 13965
Log likelihood = -162.53
-------------------------------------------------------------------------------------------
neg | Coef. Std. Err.
-------------------------------------------------------------------------------------------
left/right ideology .824** .143
Governments support for integration -.010 .007
Left/right ideology x
support for integration 1.012 * .301
Presidency .009* .002
Contributor/Beneficiary -.182* .069
Voting power .587** . 061
Number of decisions .473 .411
2000 enlargement negotiations .873 .614
Constant -2.923 1.865
---------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------
* indicates p<0.05, ** indicates p<0.01.
13
As already explained, a country can also demonstrate its dissatisfaction with a proposal through
abstaining from voting. Abstentions are effectively counted as no votes under the QMV system.
14
For an explanation please see e.g Scott Long (1997).
18
Table 7: Poisson regression Votes prior to final adoption
Random-effect Poisson regression Number of obs = 5235
Log likelihood = -101.73
--------------------------------------------------------------------------------------------
neg | Coef. Std. Err.
--------------------------------------------------------------------------------------------
left/right ideology .607** .071
Governments support for integration -.004 .003
Left/right ideology x
support for integration .713* .221
Presidency .172* .044
Contributor/Beneficiary -.228 .117
Voting power 1.028* .424
Number of decisions .473 .372
2000 enlargement negotiations .389 .209
Constant -2.017 1.728
--------------------------------------------------------------------------------------------
--------------------------------------------------------------------------------------------
* indicates p<0.05, ** indicates p<0.01.
No more than five of the variables are of significance in Table 6 and only four are
significant in Table 7 and thereby disconfirm several of the assumptions often found
in the literature. Only a governments voting power, its left/right affiliation, the
interaction term left/right ideology x support for integration and whether or not it is
holding the Presidency affect the expected number of times the government will
oppose the majority or abstain from voting when analysing votes from earlier voting
stages. A countrys contributor/beneficiary position matters only at the last voting
stage.
For the left/right ideology variable, which is significant at the 0.01 level, the
coefficients in Table 7 indicate that being a right-wing government increases the
expected number of votes against the Council majority by a factor of 1.83 (=
607 .
e )
holding all other variables constant. This result goes against the conventional
intergovernmentalist view, yet, corresponds with arguments of those scholars who
maintain that the integration dimension is no longer the only policy dimension that
matters in EU politics (Hix 1999:5). The fact that we can now see that it is not only in
the European Parliament but also in the Council that the traditional left-right
dimension plays a role further adds to this line of theory.
The interaction variable Left/right ideology x support for integration increases the
expected number of votes against the Council majority by a factor of 2.04 (=
713 .
e ) in
19
Table 7, holding everything else constant. Interestingly, the variable measuring
governments position on more/less integration did not prove significant in any of the
tables, which, together with the result from the interaction variable, means that
preferences on the integration dimension does not affect governments voting
behaviour in a simple, linear way. The interaction term shows that the effect of the
variable measuring support for integration is contingent on the governments position
on the left-right dimension.
15
As expected, countries that hold the EU Presidency oppose the majority less
frequently than other member states. The effect of this variable is 1.19 (=
172 .
e ) for
votes prior to the final adoption and 1.00 (=
009 .
e ) for the final votes, which indicates
that further research is needed to address questions such as 1) whether or not the
member state holding the Presidency do enjoy significant agenda-setting powers, and
therefore does not find it necessary to oppose the majority, or 2) whether the result
presented here reflects a wish to send a signal of political consensus, and hence makes
the country restrain itself and not vote according to its true preferences.
The Contributor/Beneficiary variable, which is at times particularly dominant in the
literature, changes from being significant at the .05 level in Table 6 to having no
noteworthy effect at all in Table 7. This could perhaps indicate that bargaining over
redistributive politics is only a matter of dispute when there is attention from the
media or the minority is facing a situation without further amendment possibilities.
In sum, and in order to make these findings more comprehensible, we can characterise
the country which will be most likely to be found in opposition to the majority as a
right-wing government possessing a large amount of voting power and which is
currently not holding the Presidency. At the final adoption stage it is furthermore
likely that this country is also a net contributor to the EU budget. However, whereas a
smaller member state would prefer to voice its opposition by use of formal statements,
this member state may equally chose to do so through voting.
Further research
All of these findings lend support to the hypotheses that a countrys left/right ideology
as well as voting power affect the member states voting behaviour and that this
15
This conclusion also corresponds with Mattilas (2004) finding, although the scores are obviously
quite different.
20
voting behaviour changes across the different stages of the legislative process. Yet,
the results are not entirely as expected: As discussed above, it would be no surprise to
find that large member states are more likely to vote against the majority than small
countries in the last possible vote. At this stage the cost of voting against the majority
is high but any opposing member state can be expected to be fully aware of this
(Crombez 2002) and makes it plausible that only certain member states are willing to
take the risk of showing their discontempt. Yet, the fact that Table 7 shows that large
member states are also more likely to vote against the majority than the smaller
member states at earlier voting stages suggests that our analysis should not end here.
Intuitively, it does not make sense that large member states are repeatedly outvoted at
the same time as voting power is an important determinant for voting outcomes.
However, I believe these result may be an expression for the following: Small states,
being rather insignificant in the broader bargaining game, may deliberately choose not
to vote against the majority at any point during the negotiations, knowing full well
that they would have to concede in the long run. The results from Table 4 and 5 show
a much greater use of formal statements rather than opposition through voting by
these countries and thereby support this argument. Small member states may instead
deliberately position themselves in a moderate position so that they can be sure to be
part of a successful coalition and then perhaps seek to influence policy outcomes
through alternative measures. Hence, what could be important bargaining techniques
for these countries are to 1) ensure an extensive use of vote trading (see also Mattila
and Lane 2001) so that smaller member states can save up the number of times they
have supported other member states (Crisp et al 2004), and in return receive
concessions on decisions that are of high saliency to them; 2) establish efficient
coodination mechanisms with parties with similar preferences in the EP as well as
within the Council itself; and releated hereto 3) perhaps even speculate in the
difference between amendment voting and final take-it-or-leave-it votes.
Therefore, all of the findings from this paper should, alongside the obvious points of
being used to update our empirical knowledge and disconfirm some important
misconceptions from the literature, also be seen as an opportunity to establish a new
research agenda, now that several important shortcomings have been identified. The
immediate tasks ahead must be to extend the analysis and explore the details of how
bargaining, vote trading and coordination takes place across policy areas, across
sectoral Councils, across the different stages in the legislative process as well as
21
across the institutions. Though, as has been pointed out throughout the paper, any
quantitative as well as qualitative study must take into account not only the last
possible votes, but also earlier voting stages as well as the extensive use of formal
statements when addressing these issues. Only then will the study of Council
decision-making be able to claim its rights as a proper political science sub-field of
EU politics.
22
References
Axelrod, R. (1970): The Conflict of Interest. Chicago: Markham.
Baldwin, R. et al. (2001): Nice Try: Should the Treaty of Nice be Ratified?
Monitoring European Integration 11, Centre for Economic Policy Research.
Banzhaf, J.F. (1965): Weighted Voting Doesnt Work: A Mathematical Analysis,
Rutgers Law Review 19:317-343.
Barr, J. and Passarelli, F (2004) Who has the Power in the EU?. Mimeo.
Best, E. (2000): The Debate over the Weighting of Votes: The Mis-Presentation of
Representation?, in Best, E. et al (eds.), Rethinking the European Union: IGC
2000 and Beyond, Maastricht: European Institute of Public Administration,
105-130.
Brams, S.J. and Affuso, P. (1985): New Paradoxes of Voting Power on the EC
Council of Ministers, Electoral Studies 4:135-139.
Budge, I. et al. (2001): Mapping Policy Preferences - Estimate for Parties, Electors,
and Governments 1945-98. Oxford, Oxford University Press.
Carruba, C.J. (1997): Net Financial Transfers in the European Union: Who gets
What and Why?, Journal of Politics 59(2):469-496.
Carruba, C.J. et al. (2004):Selection Bias in the Use of Roll Call Votes to Study
Legislative Behavior. European Parliament Research Group, Working Paper
No. 11. http://www.lse.ac.uk/Depts/eprg/working-papers.htm
Coleman, J. S. (1971): Control of Collectivities and the Power of a Collectivity to
Act in Lieberman, B. (ed.), Social Choice, New York: Gordon and Breach,
269-300.
Crisp, B. et al (2004): The Reputations Legislators Build: With Whom Should
Representatives Collaborate?. American Political Science Review (98):4.
Crombez, C. (1996): Legislative Procedures in the European Community, British
Journal of Political Science, 26, 199-228.
Crombez, C (1999): Institutional Reform and Codecision in the European Union,
http://iis-db.stanford.edu/pubs/11382/crombez.pdf.
Felsenthal, D. S. and Machover, M. (1997): The weighted voting rule in the EUs
Council of Ministers, 1958-95: intentions and outcomes. Electoral Studies
16:33-47.
Felsenthal, D. S. and Machover, M. (1998): The Measurement of Voting Power:
Theory and Practice. Cheltenham: Edward Elgar.
Gabel, M. and Huber, J.D. (2000): Putting Paties in their Place: Inferring Party Left-
Right Ideological Positions from Party Manifestos Data. American Journal of
Political Science. (44), 1, 94-103.
Galloway, D. (2001): The Treaty of Nice and Beyond: Realities and Illusions of
Power in the EU. Sheffield: Sheffield Academic Press.
Golub, J. (1999): In the Shadow of the Vote? Decision-making in the European
Community, International Organization, 53(4), 733-64.
Golub, J. (2002): Institutional Reform and Decision-Making in the European
Union, in Hosli, M.O. et al. (eds.), Institutional Challenges in the European
Union, London/New York: Routledge, 134-154.
Hagemann, S. (2005): A Council of Conflict Decision-making in the European
Unions Council of Ministers. Working Paper available upon request from
s.hagemann@lse.ac.uk.
23
Hayes-Renshaw, F. and Wallace, H. (1997): The Council of Ministers, London:
MacMillan Press Ltd.
Hix, S. (2005): The Political System of the European Union, 2nd ed. London:
Palgrave.
Hix, S. (2002): Parliamentary Behavior with Two Principals: Preferences, Parties
and Voting in the European Parliament. American Journal of Political
Science 46 (3):688-698.
Hix, S and Lord, C. (1997): Political Parties in the European Union. London:
MacMillan.
Hooghe, L. and Marks, G. (1999) The Making of a Polity: The Struggle over
European Integration Continuity and Change in Contemporary Capitalism
Kitschelt, H. et al. (eds). Cambridge: Cambridge University Press.
Holler, M. (1982) Forming Coalitions and Measuring Voting Power. Political
Studies 30: 262-271.
Hosli, M.O. (1996): Coalitions and Power: Effects of Qualified Majority Voting in
the Council of the European Union Journal of Common Market Studies
34(2):255-273.
Hosli, M.O. (1999): Power, Connected Coalitions, and Efficiency: Challenges to the
Council of the European Union International Political Science Review
20(4)371-391.
Kauppi, H. and Widgren, M. (2004): What determines EU decision making? Needs,
power or both?. Economic Policy July:221-266.
Leech, D. (2002): Designing the Voting System for the EU Council of Ministers,
Public Choice, 3-4, 437-464.
Lewis, J. (1998): Is the hard Bargaining Image of the Council Misleading? The
Committee of Permanent Representatives and the Local Elections Directive,
Journal of Common Market Studies, 36 (4):479-505.
Marks, G. and Steenbergen, M. (2002): Understanding Political Contestation in the
European Union Comparative Political Studies 35:879-892.
Marks, G. and Steenbergen, M. (2004): European Integration and Political Conflict:
Citizens, Parties, Groups. Cambridge: Cambridge University Press.
Mattila, M. (2004): Contested Decisions: Empirical analysis of voting in the
European Union Council of Ministers, European Journal of Political
Research 43:29-50.
Mattila, M. and Lane, J. E. (2001): Why Unanimity in the Council? A Roll Call
Analysis of Council Voting, European Union Politics, 2, 31-52.
Moberg, A. (1998): The Voting System in the Council of the European Union: The
Balance Between Large and Small Countries, Scandinavian Political Studies,
21, 4, 437-65.
Nugent, N. (1999): Government and politics of the European Union. Basingstoke,
Palgrave Macmillan.
Penrose, L.S. (1946): The Elementary Statistics of Majority Voting. Journal of the
Royal Statistical Society 109: 53-57.
Peters, B.G. and Wright, V (2000): The national co-ordination of EU policy : the
domestic level. New York, Oxford University Press.
Riker, W. (1962): The Theory of Political Coalitions. New Hawen & London, Yale
University Press.
Schulz, H. and Knig, T. (2000): The Efficiency of Legislative Decision Making in
The European Union, American Journal of Political Science, 4, 653-666.
Scott Long, J. (1997): Regression Models for Categorial and Limited Dependent
24
Variables. London, Sage Publications, Inc.
Steunenberg, B. et al (1999): Strategic Power in the European Union: Evaluating the
Distribution of Power in Policy Games, Journal of Theoretical Politics, 11, 3,
339-366.
Stokman et al. (2004): Winners and Losers of EU Decision Making European
Union Politics 5(1).
Taagepera, R. and Hosli, M. O. (2003): National Representation in International
Organizations: A Seat Distribution Formula for the EU Council and
Parliament (Mimeo)
Tallberg, J. (2003): The Power of the Presidency: Brokerage, Efficiency, and
Distribution in EU Negotiations. Paper presented at the EUSA International
Conference.
Tsebelis, G. and Garret, G. (1996): An institutional critique of
Intergovernmentalism. International Organisation 50(2):269-299.
Tsebelis, G. and Yataganas, X. (2002): Veto Players and Decision-making in the
EU After Nice: Policy Stability and Bureaucractic/Judicial Discretion,
Journal of Common Market Studies, 40, 2, 283-307.
Van Schendelen, M.P.C.M. (1998): EU committees as influential policy makers.
Brookfield Vt., Ashgate.
Zimmer et al. (2004) The Contested Council: The Conflict Dimension of an
Intergovernmental Institution. Paper presented at the 2nd Pan-European
Conference on EU Politics, Bologna, 24-26 June, 2004.
Minutes from the Council:
Available from:
http://ue.eu.int/cms3_applications/showPage.ASP?id=552&lang=en&mode=g
and upon request from the Secretary-General: access@consilium.eu.int.
On-line Power Index Calculation Programmes:
Bruninger, T. and Knig, T. (2001): Indices of Power (IOP), University of Konstanz,
Germany. Available at http://www.uni-konstanz.de/FuF/Verwiss/koenig/IOP.html
Pajala, A., Meskanen, T. and T. Kause (2002): POWERSLAVE Power Index Calculator: A
Voting Body Analyser in the Voting Power and Power Index Website. Published 22.4.2002.
University of Turku. Available at http://powerslave.val.utu.fi
25
Appendix 1
Distribution of votes and QMV threshold in the Council
Member
state
1958-
1972
1973-
1980
1981-
1985
1986-
1994
Since
1995
Nice
Treaty
Germany 4 10 10 10 10 29
France 10 10 10 10 29
UK 4 10 10 10 10 29
Italy 4 10 10 10 10 29
Spain 8 8 27
Poland 27
Netherlands 2 5 5 5 5 13
Greece 5 5 5 12
Belgium 2 5 5 5 5 12
Czech Rep. 12
Portugal 5 12
Hungary 12
Sweden 4 10
Austria 4 10
Slovakia 7
Denmark 3 3 3 3 7
Finland 3 7
Ireland 3 3 3 3 7
Lithuania 7
Latvia 4
Slovenia 4
Estonia 4
Cyprus 4
Luxembourg 1 2 2 2 2 4
Malta 3
Total 17 58 63 76 87 321
QMV
Threshold:
Voting
weights
12 41 45 54 62 232
Population
282.7 mio
(62%)
Member
states
13
Source: http://europa.eu.int/institutions/council/index_en.htm.
26
Appendix 2
The normalised Banzhaf index
The background for the normalised Banzhaf index is the following: A voting body has
N members with voting weights
n
w w w ,..., ,
2 1
and a decision rule in terms of a
threshold. A particular combination of votes is referred to as a division and the
Banzhaf index is essentially concerned with counting the number of swings, that is,
the voting outcomes that can be changed from losing to winning by members
changing how they cast their weighted vote. A swing for member i is a coalition
represented by a subset of members of the assembly:
i i i
S i S N S e , , , such that
q W
i
S j
j
<

e
and q W W
i
S j
j
i
> +

e
A swing is then a coalition where the total votes cast in favour of a particular decision
fall short of the threshold without those of member i, but equals or exceeds it when
member i joins. The number of swings for i is
i
q and the total number of swings is

e
=
N i
i
q q
~
. The total number of divisions (that is, the number of subsets of N) is 2
n
.
The relative Banzhaf index for member i is on this background the members relative
number of swings, denoted
i
| :

e
= =
N i
j i i i
q q q q | /
~
/ i = 1, 2, , n
It is worth noting here, that the relative Banzhaf index is in fact the normalised
version of the Penrose (1946) measure, which is an absolute measure of each
members voting power and is denoted
1
2 /

=
n
i
q t . Contrary to this, it is inherent in
the relative Banzhaf index that the sum will always come to 1 and the individual
scores therefore indicate the share of member i in the combined capacity of all
members to influence decisions. The table below compares the distribution of voting
weights with voting power according to the rules laid down in the Nice Treaty.
27
Vote and power distribution, 1999-2004
Source: Results generated by the Bruninger and Knig (2001) IOP programme.
Member
state
Population
(in mio.)
Votes
Normalised Banzhaf
Index
Germany 82.4 10 11.16
France 60.2 10 11.16
UK 60.1 10 11.16
Italy 58 10 11.16
Spain 40.2 8 9.24
Netherlands 16.2 5 5.87
Greece 10.7 5 5.87
Belgium 10.3 5 5.87
Portugal 10.1 5 5.87
Sweden 9.9 4 4.79
Austria 8.2 4 4.79
Denmark 5.4 3 3.59
Finland 5.2 3 3.59
Ireland 3.9 3 3.59
Luxembourg 0.5 2 2.26
Total 455.9 87 99.97