- Email: [email protected]
Evaluating competing criteria for allocating parliamentary seats
Mathematical Social Sciences 63 (2012) 85–89
Contents lists available at SciVerse ScienceDirect
Mathematical Social Sciences journal homepage: www.elsevier.com/locate/econbase
Evaluating competing criteria for allocating parliamentary seats✩ Richard Rose a,∗ , Patrick Bernhagen b , Gabriela Borz b a
Centre for the Study of Public Policy, University of Strathclyde, United Kingdom
b
Centre for the Study of Public Policy, University of Aberdeen, United Kingdom
article
info
Article history: Received 22 August 2011 Received in revised form 3 October 2011 Accepted 26 October 2011 Available online 9 November 2011
abstract In an established parliament any proposal for the allocation of seats will affect sitting members and their parties and is therefore likely to be evaluated by incumbents in terms of its effects on the seats that they hold. This paper evaluates the Cambridge Compromise’s formula in relation to compromises between big and small states that have characterised the EU since its foundation. It also evaluates the formula by the degree to which the Compromise departs from normative standards of equality among citizens and its distribution of seats creates more anxiety about the risks of losses as against hypothetical gains. These political criteria explain the objections to the Cambridge Compromise. However, the pressure to change the allocation of seats is continuing with EU enlargement and the arbitrary ceiling of 751 seats imposed by the Lisbon Treaty. © 2011 Elsevier B.V. All rights reserved.
European institutions have been created through treaties between states that have equality under international law regardless of population. However, gross inequalities in population between European Union (EU) member states – Germany is more than 200 times bigger than Luxembourg – have resulted in the votes assigned states in the European Council, the EU’s chief decision-making body, being unequal between legally equal states. The allocation of seats in the first chamber of national parliaments normally allocates seats to constituencies in proportion to population consistent with the norm of political equality of electors. The European Parliament (EP) has departed from the norm of the equality of Europe’s citizens by giving a disproportionate number of seats to smaller states. The result, in the words of a document of the Committee on Constitutional Affairs (2010, 39), ‘is really no more than a political fix’ involving party, national and personal interests. Moreover, there is a recurring problem of patching up the fix, because the Parliament is required to review the allocation of seats before each quinquennial EP election. In the early stages of reviewing the allocation of seats for the 2014 EP election, the EP’s Committee on Constitutional Affairs commissioned a Symposium of Mathematicians in Cambridge to identify a formula that would be ‘impartial to politics’ and ‘eliminate the political bartering which has characterised the distribution of seats so far’ (Grimmett, 2012, 1). The convenor,
✩ This paper is part of a project on Representing Europeans funded by British Economic & Social Research Council grant RES-062-23-1892. ∗ Corresponding author. E-mail address: [email protected] (R. Rose).
0165-4896/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.mathsocsci.2011.10.005
Grimmett (2012, 3), describes these terms of reference as calling for ‘a principled and fresh approach unprejudiced with respect to particular Member States or political groups and free of influence from historical positions’. The result is the aptly named Cambridge Compromise, which allocates seats taking into account both the claims of the EU’s member states (equal treatment in the Base) and population (Proportionality). The tightly focused terms of reference of the Constitutional Affairs Committee are notable for criteria that are excluded. The emphasis on a fresh approach free of historical commitments ignores positive theories that emphasise that established institutions only alter incrementally through path-dependent adjustments from a given starting point (Braybrooke and Lindblom, 1963; Pierson, 2004). Whatever the impartial intent of a formula, rational choice theories (e.g. Knight et al., 2008) predict that politicians will evaluate proposals for changes according to the anticipated effect they will have on their interests as Members of the European Parliament (MEPs) or of an EP Party Group. EU rules for approving changes in the allocation of EP seats are consistent with the Pareto optimal principle that a change should only occur if it is possible to make at least one beneficiary better off without making any others worse off, because the re-allocation of seats between countries requires the unanimous approval of member states as well as approval by a majority of MEPs. Both positive and normative theories of democracy (see e.g. Dahl, 1989) emphasise the importance of equality; individuals should not only have the right to vote but also that votes should be cast and counted on the basis of ‘one person, one vote, one value’. While political criteria were exogenous to the recommendations of the Cambridge group, they have been central in the deliberations of MEPs elected under the existing allocation of seats and
86
R. Rose et al. / Mathematical Social Sciences 63 (2012) 85–89
belonging to Party Groups that vary in the countries from which they draw their members. Instead of evaluating the apportionment of seats according to internal properties of a formula, Members of the European Parliament (MEPs) apply consequentialist criteria: How does it affect the existing allocation of seats to my country and to my Party Group? Hence, this article takes the allocation of seats in the current Parliament as the base line and makes political criteria endogenous. Since different mathematical formulas can have different consequences, the Cambridge Compromise is compared with the partisan allocation of seats with a square root formula as well as with the status quo. 1. Base + prop compromise consistent with the EU compromise 1.1. Compromising the representation of citizens and states The European Union requires all its member states to be democratic and Article 9 of the Lisbon Treaty affirms that equality of citizens is integral to the EU: ‘In all its activities, the Union shall observe the principle of equality of its citizens, who shall receive equal attention from its institutions, bodies, offices and agencies’. This commitment is reflected in the requirement that within each member state MEPs must be elected by proportional representation. However, between member states EU citizens are not equally represented in the European Parliament. Instead, the value of a citizen’s vote depends on his or her national citizenship. Article 14 of the Lisbon Treaty stipulates that seats must be allocated to member states according to the principle of degressive proportionality. The ratio between the population and the number of seats of each member state must increase in such a way that each MEP from a more populous Member State represents more citizens than each MEP from a less populous state (Lamassoure and Severin, 2007, 5). The EP’s systematic application of the principle of disproportional representation in the apportionment of seats is unique among democratically elected unicameral parliaments (Rose, 2011; cf. Pollak et al., 2011). In addition, the Lisbon Treaty imposes a basic allocation of six seats to each member state, a maximum allocation of 96 seats, and a ceiling of 751 seats in the membership of the Parliament. Of the EU’s 27 member states, 21 are over-represented in terms of a pan-European population quota for obtaining a seat. The median member state, Sweden, receives a third more MEPs than an egalitarian pan-European quota would justify and seven countries get more than double the number of MEPs they would have had if seats had been allocated strictly in proportion to population (Rose and Bernhagen, 2010). The over-representation of less populous states reflects path dependence. The first European institutions were created in the 1950s as a compact between three very populous states – France, Germany and Italy – and three relatively small states, the Benelux countries. Consistent with the equality of sovereign states in an intergovernmental institution, unanimity was initially required before the European Council could adopt significant policies. This obstacle to European integration was then replaced by assigning votes to member states with some regard to population and requiring ‘‘super majorities’’ to approve measures through complex formulas of Qualified Majority Voting (Nugent, 2010, 151ff). In the European Parliament an absolute majority is sufficient to carry a measure. The four biggest states have twofifths of the parliament’s MEPs and Germany alone has as many MEPs as the ten smallest countries. However, because MEPs vote in trans-national Party Groups rather than national blocs, the representatives of big states do not vote as a united front. The importance of big states is mediated within Party Groups that must have members from at least seven countries.
1.2. Inequality is a matter of degree The degree of inequality of citizen representation between the different member states can be measured with the Gini coefficient, a standard summary statistic of the extent of inequality (Cowell, 2011). Given that the number of citizens per MEP is ordered from the smallest to the largest, the formula is: n
G=
(2i − n − 1)x′i
i =1
n2 µ
where i is the individual country’s rank order number, n the number of countries, x′ is the country’s variable value and µ is the population average. The coefficient is 0 if there is an exactly equal distribution of seats in relation to population and 1 if the distribution is totally unequal. 1.2.1 For the EP Parliament elected in 2009, the Gini coefficient for the allocation of seats by population is 0.27. There is also a substantial measure of inequality for the 21 countries that are over-represented in their MEPs; within this group the Gini coefficient is 0.24. We can gain a sense of whether this degree of inequality is high or low by making comparisons with the bicameral Federal German Parliament and United States Congress. In each system, seats in one chamber are allocated on an egalitarian basis while in the other chamber federal partners are represented by a different formula. The apportionment of seats according to population in the American House of Representatives and in the German Bundestag results in each having a Gini coefficient of 0.04. In the United States Senate seats are allocated equally between states: each has two senators. Since there are population differences of up to 59 to 1 between American states, the Gini coefficient for the US Senate is 0.50. In the German Bundesrat, between three and six seats are allocated to its 16 Länder. Since the population differences are as great as 14 to 1, the result is a Gini coefficient of 0.35. The value of the EP’s Gini coefficient fluctuates over time as the number of member states increases. Since newer states are more often less populous, they benefit from degressive proportionality. In the 9-member Parliament elected in 1979, the coefficient was 0.21; it fell to 0.19 in 1989, when the EU had 12 member states. With the admission of 11 smaller states in 2004, the Gini coefficient rose to its current level of 0.27. Rising inequality illustrates what Hacker (2006, 401) calls the path-dependent dynamic of ‘drift within the bounds of formally stable policies’ (see also Young, 1998). 1.2.2 The Cambridge Compromise is fresh since its starting point is not the current allocation of seats between member states. Because it sets out a formula that can be applied, whatever the number of member states, it is durable, and because the formula involves generic attributes it is impartial. However, because it accepts the constraints of the EP, it is a compromise between competing principles of representation, an equal base allocation to all countries and a proportional allocation to more populous countries. The Base allocates the same number of seats, six, to each member state irrespective of its population, a total of 162 seats. This leaves a big majority, 589 seats, that can be allocated in proportion to national population up to a maximum of 96 seats for a single country, a limit stipulated in the Lisbon Treaty (Grimmett, 2012, Table 1). Allocating four-fifths of seats with some, albeit qualified, regard to population marginally reduces the Gini coefficient after rounding to 0.24.
R. Rose et al. / Mathematical Social Sciences 63 (2012) 85–89
1.2.3 A group of European academics calling themselves Scientists for a Democratic Europe (2008) endorse Penrose’s (1946) square root formula as an ‘efficient, representative, transparent and objective’ method for allocating EP seats. This method allocates seats on the basis of the square root of a country’s population. Since it applies a single formula to the allocation of all 751 seats, it meets criteria laid down by the Constitutional Affairs Committee. However, because the square root formula benefits the less populous member states it would increase inequality among European citizens and raise the Gini coefficient to 0.36 (Rose and Bernhagen, 2010, Appendix Table 5). Although inequality was not part of the terms of reference of the Constitutional Affairs Committee, the Cambridge Compromise does produce a slight reduction in the Gini coefficient by comparison with the 2009 Parliament. It is also substantially less unequal than the square root formula. Both the existing EP allocation of seats and the Cambridge Compromise are also substantially less unequal than the current allocation of votes to member states in the European Council, which has a Gini coefficient of 0.35. 2. Losers trump winners in asymmetrical churning For incumbent politicians the starting point in evaluating apportionment is the effect of re-apportioning the seats that they currently hold. A transparent formula makes the consequences of altering the allocation of seats between countries readily apparent. This is important to individual MEPs as well as parties, since the great majority of MEPs are elected in a national constituency rather than a single-member district. Re-apportionment results in churning, that is, the circulation of seats between countries and parties. With a fixed number of seats, however many seats are re-circulated among gainers and losers, the net effect is nil. However, this does not mean that the distribution of political costs and benefits is equal. The subtraction of a single seat from a country by re-apportionment makes it certain that some party will lose an MEP. This creates risk and anxiety among its MEPs and parties, since none can be certain who will bear the loss. For this reason, the uncertainty generated by the reduction of a country’s MEPs will have a multiplier effect in generating anxiety among many MEPs and parties. If more than one seat is subtracted from a country’s allocation, the anxiety generated can spread to all its parties and intensify among its MEPs too. Even though re-allocating seats creates winners in countries that gain MEPs, by definition these seats will go to politicians that are not in the European Parliament whereas losses are imposed on incumbents, and it is incumbents who have a vote on change. There is thus an asymmetry in response: losers are more likely to oppose change than winners, since the perceived losses are immediate and visible to those already advantaged, whereas the gains are potential and accrue to those who are not present. Moreover, since risk tends to be widespread the expectation of losses is magnified. For example, even if only one seat is subtracted, if four parties each calculate that they have a 50–50 chance of losing that seat, then the expected collective loss is double the actual loss.
87
The 2009 EP election awarded 736 seats to 27 member states but shortly after the election the Lisbon Treaty came into effect, authorizing a maximum of 751 MEPs. The additional 15 seats were allocated consistent with Pareto optimality: 12 countries gained seats and no country lost a seat. In consequence there is now no reserve of unallocated seats that can be drawn on to benefit some countries without imposing losses on others. Under the Lisbon Treaty any newly admitted EU member state can claim a minimum of at least six EP seats and this claim can only be met by taking seats from existing states. In short, the static consequences of Pareto optimality cannot provide a durable formula in a European Union that continues to enlarge. 2.2 European integration has progressed by promoting the norm of consensus, a vague term that avoids the liberum veto character of Pareto optimality by abandoning the requirement of unanimity (Naurin and Wallace, 2008). Consensus is a vague political term rather than a precise formula for deciding the outcome of a vote. It is evidenced by general agreement among those positive about a settlement and silent acquiescence by a limited minority that disagrees. The size of those in favour must be substantial enough to intimidate into silence those who do not want to bear the onus of disrupting a notional consensus. However, silent acquiescence is much harder to come by when losses are transparent, as in the case of a reduction in a country’s number of MEPs. Losers have strong incentives to articulate their views and to organise opposition to proposed changes. The numerical strength of those who refuse to acquiesce to the declaration of a consensus is secondary compared to the intensity of opposition, since an intense minority is sufficient to disrupt any consensus. Thus, a simple tabulation of the net total of winners is insufficient. It is also necessary to take into account whether the losers are sufficient in size to constitute an intense and vocal minority disruptive of consensus. The weight given to Prop (that is, population) in the Cambridge Compromise means that its proposed redistribution of 751 EP seats would benefit six countries, all but one among the most populous, while taking seats from 15 countries. The Czech Republic, Hungary and Portugal would lose the most MEPs, 4 each from their current allocation of 22 seats, and Belgium, Greece and Sweden would each lose 3 seats.2 Due to the fixed base of six seats per country, four of the five least populous states would retain the same number of seats. The losers are widely distributed from the states as small as Latvia (2.2 mn) to a state almost ten times as populous, Romania (21.4 mn). The proposed gains are as big as 11 seats for France and 8 each for the United Kingdom and Spain. The square root formula has the opposite effect of the Base + Prop approach. Because it favours less populous countries, it would increase the number of seats in 22 countries whilst confining losses to five countries (Rose and Bernhagen, 2010, Appendix Table 5). However, the losers would include the four biggest countries in the EU. If the square root formula were employed, Germany would suffer a loss of 29 seats, followed by Italy, the United Kingdom and France, which would collectively lose 33 seats. The countries losing seats are big enough to block any change in the EP even before a proposal reached the European Council.
2.1 3. Churning among EP parties In principle Pareto optimality does not veto all change–provided that there are additional resources to distribute. When this is the case, one country’s gain in seats would not be another’s loss.1
The mandatory use of proportional representation in EP elections reduces the partisan bias by comparison with elections
1 Of course, adding one seat to a country would alter the relative size of national delegations, but in a 751-seat parliament the effect is both trivial and diffuse.
2 Whatever formula is used, Germany will lose 3 of its 99 MEPs because the Lisbon Treaty has reduced the maximum number of seats any one country can have to 96.
88
R. Rose et al. / Mathematical Social Sciences 63 (2012) 85–89
in single-member districts. However, the removal of seats from countries with a relatively limited number of MEPs, as the Cambridge Compromise proposes, has the empirical consequence of significantly raising the threshold of votes required to gain a seat. Given that 30% of the parties electing MEPs have only a single member, their loss of one seat can have a categoric rather than a marginal impact. The outcome of re-apportionment for each Party Group need not be the same as the effect on countries or MEPs. Although the rules of the European Parliament require each Group to have MEPs elected from at least seven member states, a Group is not required to be pan-European. The number of countries represented in Party Groups ranges from 8 to 27; the median Group has members from 14 countries. Depending on the extent to which a Group draws members from all states or unevenly from larger or smaller states, churning can result in some Party Groups gaining more seats than they lose, while other Groups are net losers. 3.1 Because degressive proportionality results in the very unequal number of votes being required to elect an MEP varying between states, the number of seats won for a given share of the total European vote should vary more in Party Groups with MEPs from fewer countries. This is in fact the case (Rose and Bernhagen, 2010, Table 1). If seats in the European Parliament were distributed in exact proportion to each Party Group’s share of the total European vote, the biggest changes would occur in two smaller Groups: the Greens would gain five seats and the Conservative and Reform Group would lose 16 seats. The three largest groups, the People’s Party, the Socialists and Liberals, would each gain only one seat. 3.2 The terms of reference of the Cambridge Compromise are impartial but the absence of intent does not mean that they are without partisan effects, since there is no a priori reason to expect the effects of re-apportionment to be neutral for each multinational Party Group. We estimate the impact of the Base + Prop formula on Groups by applying the share of the national vote gained by each party at the 2009 EP election to the reduced or increased number of MEPs that the country would receive under the Compromise, and rounding off to the nearest whole number. The effect of re-apportionment on a multi-national Party Group is the sum of its effects on up to 27 countries. The result is an estimate subject to a ceteris paribus clause, and has a margin of error; such uncertainties are a stimulus to increased anxiety over the risk of losing seats (Kahnemann and Tversky, 1979). Re-apportionment according to the Cambridge Compromise would take 35 seats from fifteen countries. We estimate that 16 People’s Party MEPs would lose their seats and 13 Socialist Party members. Together, these two Party Groups have a sufficient majority in the European Parliament to defeat any proposal for change that they view as harmful to their interests. Except for the Greens, all the other Groups would also lose at least one MEP. The ‘‘Prop’’ portion of the Cambridge Compromise benefits populous countries where the biggest EP Groups are strong. The People’s Party would gain an estimated 10 seats, the Socialists 8 and the Liberals 6. Three of the smaller Groups – the Conservative and Reform, Left-Green and Freedom and Democracy – would each gain 3 seats and the Greens 2. The net effect of churning would still leave the People’s Party and the Socialists worse off by 6 and 5 seats respectively, and smaller parties would usually have small net gains.
3.3 The Square Root formula has a much higher volume of churning, affecting 71 seats in 26 countries. More than two-thirds of seats would be lost by the three biggest parties: the People’s Party would lose 26 seats, the Socialists 17 seats, and the Liberals 8 seats. In addition, the Greens and the Conservative and Reform Group would lose 7 and 6 seats respectively. The net effect would be different, for there would be an estimated gain of 5 seats by the Liberals and 4 seats by the non-aligned collection of MEPs, and the Socialists would also gain 2 seats. The biggest net losers would be the People’s Party, down 5 seats and the Conservative and Reform Group, down 5 seats. Whatever the formula applied to re-apportion seats, the arithmetic of coalition-building in the European Parliament would remain the same. An absolute majority of votes would require a ‘‘black–red’’ agreement between the two largest Groups, the European People’s Party and the Socialists or, alternatively; a coalition led by one of these parties that included a host of smaller parties; or a super-majority including the three largest parties plus other MEPs (cf. Votewatch, 2011). However many parties were in the majority coalition, there would remain a large enough minority to disrupt a consensus. 4. Political criteria are decisive Comparing the Cambridge Compromise, the square root formula, and the existing EP allocation highlights distinctive properties of each. The square root method is a pure mathematical formula. By contrast, the Cambridge proposal is a consciously chosen compromise of equality between member states (the Base) and a degree of proportionality as regards population (the Prop). The current EP allocation of seats is a path-dependent compromise, having been arrived at over more than three decades of bargains between incumbents. As far as treating citizens equally is concerned, the Cambridge Compromise is marginally better, having a Gini coefficient of 0.24 compared to 0.27 for the status quo, while the square root formula raises the Gini coefficient to 0.36. The square root formula is completely monotonic in reducing the population required for an MEP as the population of countries falls and the Base + Prop formula involves only two departures from being monotonic. Both are a substantial improvement over the 2009 pattern, in which six countries are inconsistent with this principle. In terms of national impact, the Cambridge Compromise scores worst, for it takes seats from 15 countries, whereas the square root formula takes seats from only 5 countries. By definition, the status quo is Pareto optimal, since there are no changes. The partisan impact of the square root formula is greatest, for 71 MEPs would lose their seats, as against 35 MEPs losing seats with the Cambridge Compromise. Notwithstanding the political resistance to any re-allocation of seats, given that there are no longer any unallocated seats admitting more countries to the EU necessarily requires re-allocation. With the exception of Turkey, for whom membership is hardly imminent, all the countries negotiating membership are less populous states that would benefit from degressive proportionality. The Cambridge Compromise calculates that Croatia, now scheduled to join before the 2014 EP election, would receive 11 seats at the cost of taking one MEP from each of nine countries and two from France (Grimmett, 2012, Table 2). The Gini coefficient would remain at 0.24. If Iceland were to join and claim its base allocation of six seats, this would increase the number of states losing seats to 12, with losses concentrated among the largest states. The Gini coefficient would increase to 0.27. A proposal to elect 25 MEPs from a panEuropean list, intended to ‘‘Europeanise’’ elections currently fought in national constituencies (cf. Duff, 2010, 55–63; Bardi et al., 2010), would have a bigger displacement effect on the existing distribution of seats.
R. Rose et al. / Mathematical Social Sciences 63 (2012) 85–89
Only if the ceiling of 751 EP seats is raised in a subsequent treaty revision would it become possible for re-apportionment to avoid the political consequences of asymmetrical churning creating more anxiety among states losing seats than perceived benefit among states gaining seats. The existing rigid restriction on seats could be eased by a formula that recognises enlargement. For example, the ceiling of 751 seats could be increased by taking into account the increase in EU population that results from enlargement as well as the requirement of degressive proportionality. Additionally or alternatively, a five percent reduction in the panEuropean average population per MEP would provide sufficient additional seats to produce the redistribution of seats in favour of big countries that is a consequence of the Cambridge Compromise. If the ceiling on seats were marginally raised, then political and mathematical criteria would no longer be in opposition but could be joined because giving more seats to some new member states would not generate the political uncertainty, anxiety and opposition resulting from existing practices. References Bardi, Luciano, Bressanelli, E., Calossi, E., Gagatek, W., Mair, P., Pizzimenti, E., 2010. How to create a transnational party system. In: EUDO, Robert Schuman Centre. European University Institute. Florence. Braybrooke, David, Lindblom, C.E., 1963. A Strategy of Decision. Free Press, New York. Committee on Constitutional Affairs, 2010. Draft report on a proposal for a modification of the act concerning the election of the members of the European parliament. European Parliament 2010/XXXX(INI) 12 April. Brussels. Cowell, Frank A., 2011. Measuring Inequality, 3rd ed. Oxford University Press, Oxford. Dahl, Robert A., 1989. Democracy and its Critics. Yale University Press, New Haven.
89
Duff, Andrew, 2010. Post-National Democracy and the Reform of the European Parliament. Notre Europe., Paris. Grimmett, Geoffrey, 2012. European apportionment via the Cambridge Compromise. Mathematical Social Sciences 63 (2), 68–73. Hacker, Jacob, 2006. The welfare state. In: Rhodes, R.A.W., Binder, S.A., Rockman, B.A. (Eds.), The Oxford Handbook of Political Institutions. Oxford University Press, Oxford, pp. 385–406. Kahnemann, Daniel, Tversky, Amos, 1979. Prospect theory: an analysis of decision under risk. Econometrica 47, 263–291. Knight, J., Levi, M., Johnson, J., Stokes, S., 2008. Designing Democratic Government. Russell Sage Foundation, New York. Lamassoure, Alain, Severin, Adrian, 2007. Report on the composition of the European parliament (2007/2169(INI)). Committee on Constitutional Affairs, European Parliament 2007/A6-0351(FINAL) 3 October. Brussels. Naurin, D., Helen, Wallace (Eds.), 2008. Unveiling the Council of the European Union. Palgrave Macmillan, Basingstoke. Nugent, Neil, 2010. The Government and Politics of the European Union. Palgrave Macmillan, Basingstoke. Penrose, L.S., 1946. The elementary statistics of majority voting. Journal of the Royal Statistical Society 109, 53–57. Pierson, Paul, 2004. Politics in Time: History, Institutions and Social Analysis. Princeton University Press, Princeton. Pollak, Johannes, Everson, Michelle, Lord, Christopher, Schubert, Samuel, de Wilde, Pieter, 2011. Citizens’ weight of vote in selected federal systems. European Parliament Directorate-General for Internal Policies Policy Department C. Brussels. Rose, Richard, 2011. Representation in parliamentary democracies: the European parliament as a deviant case. Centre for the Study of Public Policy. Studies in Public Policy No. 486. Aberdeen. Rose, Richard, Bernhagen, Patrick, 2010. Inequality in the representation of citizens in the European parliament. Centre for the Study of Public Policy Studies in Public Policy No. 472. Aberdeen. Scientists for a Democratic Europe, 2008. Letter to the Governments of EU Member States. Available at: www://sanver.bilgi.tr/pdf/Openletter/pdf. Votewatch,, 2011. Brussels. www.votewatch.eu/ex_epg_coalitions.php (accessed 21.04.11). Young, Peyton, 1998. Individual Strategy and Social Structure: An Evolutionary Theory of Institutions. Princeton University Press, Princeton.
Contents lists available at SciVerse ScienceDirect
Mathematical Social Sciences journal homepage: www.elsevier.com/locate/econbase
Evaluating competing criteria for allocating parliamentary seats✩ Richard Rose a,∗ , Patrick Bernhagen b , Gabriela Borz b a
Centre for the Study of Public Policy, University of Strathclyde, United Kingdom
b
Centre for the Study of Public Policy, University of Aberdeen, United Kingdom
article
info
Article history: Received 22 August 2011 Received in revised form 3 October 2011 Accepted 26 October 2011 Available online 9 November 2011
abstract In an established parliament any proposal for the allocation of seats will affect sitting members and their parties and is therefore likely to be evaluated by incumbents in terms of its effects on the seats that they hold. This paper evaluates the Cambridge Compromise’s formula in relation to compromises between big and small states that have characterised the EU since its foundation. It also evaluates the formula by the degree to which the Compromise departs from normative standards of equality among citizens and its distribution of seats creates more anxiety about the risks of losses as against hypothetical gains. These political criteria explain the objections to the Cambridge Compromise. However, the pressure to change the allocation of seats is continuing with EU enlargement and the arbitrary ceiling of 751 seats imposed by the Lisbon Treaty. © 2011 Elsevier B.V. All rights reserved.
European institutions have been created through treaties between states that have equality under international law regardless of population. However, gross inequalities in population between European Union (EU) member states – Germany is more than 200 times bigger than Luxembourg – have resulted in the votes assigned states in the European Council, the EU’s chief decision-making body, being unequal between legally equal states. The allocation of seats in the first chamber of national parliaments normally allocates seats to constituencies in proportion to population consistent with the norm of political equality of electors. The European Parliament (EP) has departed from the norm of the equality of Europe’s citizens by giving a disproportionate number of seats to smaller states. The result, in the words of a document of the Committee on Constitutional Affairs (2010, 39), ‘is really no more than a political fix’ involving party, national and personal interests. Moreover, there is a recurring problem of patching up the fix, because the Parliament is required to review the allocation of seats before each quinquennial EP election. In the early stages of reviewing the allocation of seats for the 2014 EP election, the EP’s Committee on Constitutional Affairs commissioned a Symposium of Mathematicians in Cambridge to identify a formula that would be ‘impartial to politics’ and ‘eliminate the political bartering which has characterised the distribution of seats so far’ (Grimmett, 2012, 1). The convenor,
✩ This paper is part of a project on Representing Europeans funded by British Economic & Social Research Council grant RES-062-23-1892. ∗ Corresponding author. E-mail address: [email protected] (R. Rose).
0165-4896/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.mathsocsci.2011.10.005
Grimmett (2012, 3), describes these terms of reference as calling for ‘a principled and fresh approach unprejudiced with respect to particular Member States or political groups and free of influence from historical positions’. The result is the aptly named Cambridge Compromise, which allocates seats taking into account both the claims of the EU’s member states (equal treatment in the Base) and population (Proportionality). The tightly focused terms of reference of the Constitutional Affairs Committee are notable for criteria that are excluded. The emphasis on a fresh approach free of historical commitments ignores positive theories that emphasise that established institutions only alter incrementally through path-dependent adjustments from a given starting point (Braybrooke and Lindblom, 1963; Pierson, 2004). Whatever the impartial intent of a formula, rational choice theories (e.g. Knight et al., 2008) predict that politicians will evaluate proposals for changes according to the anticipated effect they will have on their interests as Members of the European Parliament (MEPs) or of an EP Party Group. EU rules for approving changes in the allocation of EP seats are consistent with the Pareto optimal principle that a change should only occur if it is possible to make at least one beneficiary better off without making any others worse off, because the re-allocation of seats between countries requires the unanimous approval of member states as well as approval by a majority of MEPs. Both positive and normative theories of democracy (see e.g. Dahl, 1989) emphasise the importance of equality; individuals should not only have the right to vote but also that votes should be cast and counted on the basis of ‘one person, one vote, one value’. While political criteria were exogenous to the recommendations of the Cambridge group, they have been central in the deliberations of MEPs elected under the existing allocation of seats and
86
R. Rose et al. / Mathematical Social Sciences 63 (2012) 85–89
belonging to Party Groups that vary in the countries from which they draw their members. Instead of evaluating the apportionment of seats according to internal properties of a formula, Members of the European Parliament (MEPs) apply consequentialist criteria: How does it affect the existing allocation of seats to my country and to my Party Group? Hence, this article takes the allocation of seats in the current Parliament as the base line and makes political criteria endogenous. Since different mathematical formulas can have different consequences, the Cambridge Compromise is compared with the partisan allocation of seats with a square root formula as well as with the status quo. 1. Base + prop compromise consistent with the EU compromise 1.1. Compromising the representation of citizens and states The European Union requires all its member states to be democratic and Article 9 of the Lisbon Treaty affirms that equality of citizens is integral to the EU: ‘In all its activities, the Union shall observe the principle of equality of its citizens, who shall receive equal attention from its institutions, bodies, offices and agencies’. This commitment is reflected in the requirement that within each member state MEPs must be elected by proportional representation. However, between member states EU citizens are not equally represented in the European Parliament. Instead, the value of a citizen’s vote depends on his or her national citizenship. Article 14 of the Lisbon Treaty stipulates that seats must be allocated to member states according to the principle of degressive proportionality. The ratio between the population and the number of seats of each member state must increase in such a way that each MEP from a more populous Member State represents more citizens than each MEP from a less populous state (Lamassoure and Severin, 2007, 5). The EP’s systematic application of the principle of disproportional representation in the apportionment of seats is unique among democratically elected unicameral parliaments (Rose, 2011; cf. Pollak et al., 2011). In addition, the Lisbon Treaty imposes a basic allocation of six seats to each member state, a maximum allocation of 96 seats, and a ceiling of 751 seats in the membership of the Parliament. Of the EU’s 27 member states, 21 are over-represented in terms of a pan-European population quota for obtaining a seat. The median member state, Sweden, receives a third more MEPs than an egalitarian pan-European quota would justify and seven countries get more than double the number of MEPs they would have had if seats had been allocated strictly in proportion to population (Rose and Bernhagen, 2010). The over-representation of less populous states reflects path dependence. The first European institutions were created in the 1950s as a compact between three very populous states – France, Germany and Italy – and three relatively small states, the Benelux countries. Consistent with the equality of sovereign states in an intergovernmental institution, unanimity was initially required before the European Council could adopt significant policies. This obstacle to European integration was then replaced by assigning votes to member states with some regard to population and requiring ‘‘super majorities’’ to approve measures through complex formulas of Qualified Majority Voting (Nugent, 2010, 151ff). In the European Parliament an absolute majority is sufficient to carry a measure. The four biggest states have twofifths of the parliament’s MEPs and Germany alone has as many MEPs as the ten smallest countries. However, because MEPs vote in trans-national Party Groups rather than national blocs, the representatives of big states do not vote as a united front. The importance of big states is mediated within Party Groups that must have members from at least seven countries.
1.2. Inequality is a matter of degree The degree of inequality of citizen representation between the different member states can be measured with the Gini coefficient, a standard summary statistic of the extent of inequality (Cowell, 2011). Given that the number of citizens per MEP is ordered from the smallest to the largest, the formula is: n
G=
(2i − n − 1)x′i
i =1
n2 µ
where i is the individual country’s rank order number, n the number of countries, x′ is the country’s variable value and µ is the population average. The coefficient is 0 if there is an exactly equal distribution of seats in relation to population and 1 if the distribution is totally unequal. 1.2.1 For the EP Parliament elected in 2009, the Gini coefficient for the allocation of seats by population is 0.27. There is also a substantial measure of inequality for the 21 countries that are over-represented in their MEPs; within this group the Gini coefficient is 0.24. We can gain a sense of whether this degree of inequality is high or low by making comparisons with the bicameral Federal German Parliament and United States Congress. In each system, seats in one chamber are allocated on an egalitarian basis while in the other chamber federal partners are represented by a different formula. The apportionment of seats according to population in the American House of Representatives and in the German Bundestag results in each having a Gini coefficient of 0.04. In the United States Senate seats are allocated equally between states: each has two senators. Since there are population differences of up to 59 to 1 between American states, the Gini coefficient for the US Senate is 0.50. In the German Bundesrat, between three and six seats are allocated to its 16 Länder. Since the population differences are as great as 14 to 1, the result is a Gini coefficient of 0.35. The value of the EP’s Gini coefficient fluctuates over time as the number of member states increases. Since newer states are more often less populous, they benefit from degressive proportionality. In the 9-member Parliament elected in 1979, the coefficient was 0.21; it fell to 0.19 in 1989, when the EU had 12 member states. With the admission of 11 smaller states in 2004, the Gini coefficient rose to its current level of 0.27. Rising inequality illustrates what Hacker (2006, 401) calls the path-dependent dynamic of ‘drift within the bounds of formally stable policies’ (see also Young, 1998). 1.2.2 The Cambridge Compromise is fresh since its starting point is not the current allocation of seats between member states. Because it sets out a formula that can be applied, whatever the number of member states, it is durable, and because the formula involves generic attributes it is impartial. However, because it accepts the constraints of the EP, it is a compromise between competing principles of representation, an equal base allocation to all countries and a proportional allocation to more populous countries. The Base allocates the same number of seats, six, to each member state irrespective of its population, a total of 162 seats. This leaves a big majority, 589 seats, that can be allocated in proportion to national population up to a maximum of 96 seats for a single country, a limit stipulated in the Lisbon Treaty (Grimmett, 2012, Table 1). Allocating four-fifths of seats with some, albeit qualified, regard to population marginally reduces the Gini coefficient after rounding to 0.24.
R. Rose et al. / Mathematical Social Sciences 63 (2012) 85–89
1.2.3 A group of European academics calling themselves Scientists for a Democratic Europe (2008) endorse Penrose’s (1946) square root formula as an ‘efficient, representative, transparent and objective’ method for allocating EP seats. This method allocates seats on the basis of the square root of a country’s population. Since it applies a single formula to the allocation of all 751 seats, it meets criteria laid down by the Constitutional Affairs Committee. However, because the square root formula benefits the less populous member states it would increase inequality among European citizens and raise the Gini coefficient to 0.36 (Rose and Bernhagen, 2010, Appendix Table 5). Although inequality was not part of the terms of reference of the Constitutional Affairs Committee, the Cambridge Compromise does produce a slight reduction in the Gini coefficient by comparison with the 2009 Parliament. It is also substantially less unequal than the square root formula. Both the existing EP allocation of seats and the Cambridge Compromise are also substantially less unequal than the current allocation of votes to member states in the European Council, which has a Gini coefficient of 0.35. 2. Losers trump winners in asymmetrical churning For incumbent politicians the starting point in evaluating apportionment is the effect of re-apportioning the seats that they currently hold. A transparent formula makes the consequences of altering the allocation of seats between countries readily apparent. This is important to individual MEPs as well as parties, since the great majority of MEPs are elected in a national constituency rather than a single-member district. Re-apportionment results in churning, that is, the circulation of seats between countries and parties. With a fixed number of seats, however many seats are re-circulated among gainers and losers, the net effect is nil. However, this does not mean that the distribution of political costs and benefits is equal. The subtraction of a single seat from a country by re-apportionment makes it certain that some party will lose an MEP. This creates risk and anxiety among its MEPs and parties, since none can be certain who will bear the loss. For this reason, the uncertainty generated by the reduction of a country’s MEPs will have a multiplier effect in generating anxiety among many MEPs and parties. If more than one seat is subtracted from a country’s allocation, the anxiety generated can spread to all its parties and intensify among its MEPs too. Even though re-allocating seats creates winners in countries that gain MEPs, by definition these seats will go to politicians that are not in the European Parliament whereas losses are imposed on incumbents, and it is incumbents who have a vote on change. There is thus an asymmetry in response: losers are more likely to oppose change than winners, since the perceived losses are immediate and visible to those already advantaged, whereas the gains are potential and accrue to those who are not present. Moreover, since risk tends to be widespread the expectation of losses is magnified. For example, even if only one seat is subtracted, if four parties each calculate that they have a 50–50 chance of losing that seat, then the expected collective loss is double the actual loss.
87
The 2009 EP election awarded 736 seats to 27 member states but shortly after the election the Lisbon Treaty came into effect, authorizing a maximum of 751 MEPs. The additional 15 seats were allocated consistent with Pareto optimality: 12 countries gained seats and no country lost a seat. In consequence there is now no reserve of unallocated seats that can be drawn on to benefit some countries without imposing losses on others. Under the Lisbon Treaty any newly admitted EU member state can claim a minimum of at least six EP seats and this claim can only be met by taking seats from existing states. In short, the static consequences of Pareto optimality cannot provide a durable formula in a European Union that continues to enlarge. 2.2 European integration has progressed by promoting the norm of consensus, a vague term that avoids the liberum veto character of Pareto optimality by abandoning the requirement of unanimity (Naurin and Wallace, 2008). Consensus is a vague political term rather than a precise formula for deciding the outcome of a vote. It is evidenced by general agreement among those positive about a settlement and silent acquiescence by a limited minority that disagrees. The size of those in favour must be substantial enough to intimidate into silence those who do not want to bear the onus of disrupting a notional consensus. However, silent acquiescence is much harder to come by when losses are transparent, as in the case of a reduction in a country’s number of MEPs. Losers have strong incentives to articulate their views and to organise opposition to proposed changes. The numerical strength of those who refuse to acquiesce to the declaration of a consensus is secondary compared to the intensity of opposition, since an intense minority is sufficient to disrupt any consensus. Thus, a simple tabulation of the net total of winners is insufficient. It is also necessary to take into account whether the losers are sufficient in size to constitute an intense and vocal minority disruptive of consensus. The weight given to Prop (that is, population) in the Cambridge Compromise means that its proposed redistribution of 751 EP seats would benefit six countries, all but one among the most populous, while taking seats from 15 countries. The Czech Republic, Hungary and Portugal would lose the most MEPs, 4 each from their current allocation of 22 seats, and Belgium, Greece and Sweden would each lose 3 seats.2 Due to the fixed base of six seats per country, four of the five least populous states would retain the same number of seats. The losers are widely distributed from the states as small as Latvia (2.2 mn) to a state almost ten times as populous, Romania (21.4 mn). The proposed gains are as big as 11 seats for France and 8 each for the United Kingdom and Spain. The square root formula has the opposite effect of the Base + Prop approach. Because it favours less populous countries, it would increase the number of seats in 22 countries whilst confining losses to five countries (Rose and Bernhagen, 2010, Appendix Table 5). However, the losers would include the four biggest countries in the EU. If the square root formula were employed, Germany would suffer a loss of 29 seats, followed by Italy, the United Kingdom and France, which would collectively lose 33 seats. The countries losing seats are big enough to block any change in the EP even before a proposal reached the European Council.
2.1 3. Churning among EP parties In principle Pareto optimality does not veto all change–provided that there are additional resources to distribute. When this is the case, one country’s gain in seats would not be another’s loss.1
The mandatory use of proportional representation in EP elections reduces the partisan bias by comparison with elections
1 Of course, adding one seat to a country would alter the relative size of national delegations, but in a 751-seat parliament the effect is both trivial and diffuse.
2 Whatever formula is used, Germany will lose 3 of its 99 MEPs because the Lisbon Treaty has reduced the maximum number of seats any one country can have to 96.
88
R. Rose et al. / Mathematical Social Sciences 63 (2012) 85–89
in single-member districts. However, the removal of seats from countries with a relatively limited number of MEPs, as the Cambridge Compromise proposes, has the empirical consequence of significantly raising the threshold of votes required to gain a seat. Given that 30% of the parties electing MEPs have only a single member, their loss of one seat can have a categoric rather than a marginal impact. The outcome of re-apportionment for each Party Group need not be the same as the effect on countries or MEPs. Although the rules of the European Parliament require each Group to have MEPs elected from at least seven member states, a Group is not required to be pan-European. The number of countries represented in Party Groups ranges from 8 to 27; the median Group has members from 14 countries. Depending on the extent to which a Group draws members from all states or unevenly from larger or smaller states, churning can result in some Party Groups gaining more seats than they lose, while other Groups are net losers. 3.1 Because degressive proportionality results in the very unequal number of votes being required to elect an MEP varying between states, the number of seats won for a given share of the total European vote should vary more in Party Groups with MEPs from fewer countries. This is in fact the case (Rose and Bernhagen, 2010, Table 1). If seats in the European Parliament were distributed in exact proportion to each Party Group’s share of the total European vote, the biggest changes would occur in two smaller Groups: the Greens would gain five seats and the Conservative and Reform Group would lose 16 seats. The three largest groups, the People’s Party, the Socialists and Liberals, would each gain only one seat. 3.2 The terms of reference of the Cambridge Compromise are impartial but the absence of intent does not mean that they are without partisan effects, since there is no a priori reason to expect the effects of re-apportionment to be neutral for each multinational Party Group. We estimate the impact of the Base + Prop formula on Groups by applying the share of the national vote gained by each party at the 2009 EP election to the reduced or increased number of MEPs that the country would receive under the Compromise, and rounding off to the nearest whole number. The effect of re-apportionment on a multi-national Party Group is the sum of its effects on up to 27 countries. The result is an estimate subject to a ceteris paribus clause, and has a margin of error; such uncertainties are a stimulus to increased anxiety over the risk of losing seats (Kahnemann and Tversky, 1979). Re-apportionment according to the Cambridge Compromise would take 35 seats from fifteen countries. We estimate that 16 People’s Party MEPs would lose their seats and 13 Socialist Party members. Together, these two Party Groups have a sufficient majority in the European Parliament to defeat any proposal for change that they view as harmful to their interests. Except for the Greens, all the other Groups would also lose at least one MEP. The ‘‘Prop’’ portion of the Cambridge Compromise benefits populous countries where the biggest EP Groups are strong. The People’s Party would gain an estimated 10 seats, the Socialists 8 and the Liberals 6. Three of the smaller Groups – the Conservative and Reform, Left-Green and Freedom and Democracy – would each gain 3 seats and the Greens 2. The net effect of churning would still leave the People’s Party and the Socialists worse off by 6 and 5 seats respectively, and smaller parties would usually have small net gains.
3.3 The Square Root formula has a much higher volume of churning, affecting 71 seats in 26 countries. More than two-thirds of seats would be lost by the three biggest parties: the People’s Party would lose 26 seats, the Socialists 17 seats, and the Liberals 8 seats. In addition, the Greens and the Conservative and Reform Group would lose 7 and 6 seats respectively. The net effect would be different, for there would be an estimated gain of 5 seats by the Liberals and 4 seats by the non-aligned collection of MEPs, and the Socialists would also gain 2 seats. The biggest net losers would be the People’s Party, down 5 seats and the Conservative and Reform Group, down 5 seats. Whatever the formula applied to re-apportion seats, the arithmetic of coalition-building in the European Parliament would remain the same. An absolute majority of votes would require a ‘‘black–red’’ agreement between the two largest Groups, the European People’s Party and the Socialists or, alternatively; a coalition led by one of these parties that included a host of smaller parties; or a super-majority including the three largest parties plus other MEPs (cf. Votewatch, 2011). However many parties were in the majority coalition, there would remain a large enough minority to disrupt a consensus. 4. Political criteria are decisive Comparing the Cambridge Compromise, the square root formula, and the existing EP allocation highlights distinctive properties of each. The square root method is a pure mathematical formula. By contrast, the Cambridge proposal is a consciously chosen compromise of equality between member states (the Base) and a degree of proportionality as regards population (the Prop). The current EP allocation of seats is a path-dependent compromise, having been arrived at over more than three decades of bargains between incumbents. As far as treating citizens equally is concerned, the Cambridge Compromise is marginally better, having a Gini coefficient of 0.24 compared to 0.27 for the status quo, while the square root formula raises the Gini coefficient to 0.36. The square root formula is completely monotonic in reducing the population required for an MEP as the population of countries falls and the Base + Prop formula involves only two departures from being monotonic. Both are a substantial improvement over the 2009 pattern, in which six countries are inconsistent with this principle. In terms of national impact, the Cambridge Compromise scores worst, for it takes seats from 15 countries, whereas the square root formula takes seats from only 5 countries. By definition, the status quo is Pareto optimal, since there are no changes. The partisan impact of the square root formula is greatest, for 71 MEPs would lose their seats, as against 35 MEPs losing seats with the Cambridge Compromise. Notwithstanding the political resistance to any re-allocation of seats, given that there are no longer any unallocated seats admitting more countries to the EU necessarily requires re-allocation. With the exception of Turkey, for whom membership is hardly imminent, all the countries negotiating membership are less populous states that would benefit from degressive proportionality. The Cambridge Compromise calculates that Croatia, now scheduled to join before the 2014 EP election, would receive 11 seats at the cost of taking one MEP from each of nine countries and two from France (Grimmett, 2012, Table 2). The Gini coefficient would remain at 0.24. If Iceland were to join and claim its base allocation of six seats, this would increase the number of states losing seats to 12, with losses concentrated among the largest states. The Gini coefficient would increase to 0.27. A proposal to elect 25 MEPs from a panEuropean list, intended to ‘‘Europeanise’’ elections currently fought in national constituencies (cf. Duff, 2010, 55–63; Bardi et al., 2010), would have a bigger displacement effect on the existing distribution of seats.
R. Rose et al. / Mathematical Social Sciences 63 (2012) 85–89
Only if the ceiling of 751 EP seats is raised in a subsequent treaty revision would it become possible for re-apportionment to avoid the political consequences of asymmetrical churning creating more anxiety among states losing seats than perceived benefit among states gaining seats. The existing rigid restriction on seats could be eased by a formula that recognises enlargement. For example, the ceiling of 751 seats could be increased by taking into account the increase in EU population that results from enlargement as well as the requirement of degressive proportionality. Additionally or alternatively, a five percent reduction in the panEuropean average population per MEP would provide sufficient additional seats to produce the redistribution of seats in favour of big countries that is a consequence of the Cambridge Compromise. If the ceiling on seats were marginally raised, then political and mathematical criteria would no longer be in opposition but could be joined because giving more seats to some new member states would not generate the political uncertainty, anxiety and opposition resulting from existing practices. References Bardi, Luciano, Bressanelli, E., Calossi, E., Gagatek, W., Mair, P., Pizzimenti, E., 2010. How to create a transnational party system. In: EUDO, Robert Schuman Centre. European University Institute. Florence. Braybrooke, David, Lindblom, C.E., 1963. A Strategy of Decision. Free Press, New York. Committee on Constitutional Affairs, 2010. Draft report on a proposal for a modification of the act concerning the election of the members of the European parliament. European Parliament 2010/XXXX(INI) 12 April. Brussels. Cowell, Frank A., 2011. Measuring Inequality, 3rd ed. Oxford University Press, Oxford. Dahl, Robert A., 1989. Democracy and its Critics. Yale University Press, New Haven.
89
Duff, Andrew, 2010. Post-National Democracy and the Reform of the European Parliament. Notre Europe., Paris. Grimmett, Geoffrey, 2012. European apportionment via the Cambridge Compromise. Mathematical Social Sciences 63 (2), 68–73. Hacker, Jacob, 2006. The welfare state. In: Rhodes, R.A.W., Binder, S.A., Rockman, B.A. (Eds.), The Oxford Handbook of Political Institutions. Oxford University Press, Oxford, pp. 385–406. Kahnemann, Daniel, Tversky, Amos, 1979. Prospect theory: an analysis of decision under risk. Econometrica 47, 263–291. Knight, J., Levi, M., Johnson, J., Stokes, S., 2008. Designing Democratic Government. Russell Sage Foundation, New York. Lamassoure, Alain, Severin, Adrian, 2007. Report on the composition of the European parliament (2007/2169(INI)). Committee on Constitutional Affairs, European Parliament 2007/A6-0351(FINAL) 3 October. Brussels. Naurin, D., Helen, Wallace (Eds.), 2008. Unveiling the Council of the European Union. Palgrave Macmillan, Basingstoke. Nugent, Neil, 2010. The Government and Politics of the European Union. Palgrave Macmillan, Basingstoke. Penrose, L.S., 1946. The elementary statistics of majority voting. Journal of the Royal Statistical Society 109, 53–57. Pierson, Paul, 2004. Politics in Time: History, Institutions and Social Analysis. Princeton University Press, Princeton. Pollak, Johannes, Everson, Michelle, Lord, Christopher, Schubert, Samuel, de Wilde, Pieter, 2011. Citizens’ weight of vote in selected federal systems. European Parliament Directorate-General for Internal Policies Policy Department C. Brussels. Rose, Richard, 2011. Representation in parliamentary democracies: the European parliament as a deviant case. Centre for the Study of Public Policy. Studies in Public Policy No. 486. Aberdeen. Rose, Richard, Bernhagen, Patrick, 2010. Inequality in the representation of citizens in the European parliament. Centre for the Study of Public Policy Studies in Public Policy No. 472. Aberdeen. Scientists for a Democratic Europe, 2008. Letter to the Governments of EU Member States. Available at: www://sanver.bilgi.tr/pdf/Openletter/pdf. Votewatch,, 2011. Brussels. www.votewatch.eu/ex_epg_coalitions.php (accessed 21.04.11). Young, Peyton, 1998. Individual Strategy and Social Structure: An Evolutionary Theory of Institutions. Princeton University Press, Princeton.