Improving policy representation of the German Bundestag by an alternative election method. Part I: Analysis

Transcription

1 Improvin policy representation of the German Bndesta by an alternative election method. art I: Analysis Andranik Tanian Institte of Economic and Social Research (WSI) at the Hans Böckler Fondation, Düsseldorf, and Karlsrhe Institte of Technoloy, Karlsrhe, Germany April 14, 2016 Abstract Since voters are often swayed more by the charisma, personal imae and commnication skills of the individal candidates standin for election than by the parties political manifestos, they may cast votes that are actally in opposition to their policy preferences. Sch a type of behavior, known as irrational votin, reslts in the election of representatives who do not correspond exactly to the voters own views. This is illstrated in art I of this paper, Analysis, sin the example of 2013 German Bndesta (federal parliament). We consider the 28 parties that took part in the 2013 Bndesta election and compare their positions on 36 topical isses with the reslts of pblic opinion polls. Then we constrct the party and coalition indices of poplarity (the averae percentae of the poplation represented) and niversality (freqency in representin a majority). In particlar, we find that the 2013 election winner, the conservative nion CD/CS with their 41.5% of the votes, was the least representative amon the 28 parties considered. The representativeness of the Bndesta is abot 50%, as if the correspondence with the electorate s preference on every policy isse had been decided by tossin a coin, meanin that the Bndesta s policy profile is independent of that of the electorate. In the second part of the paper, roposal, an alternative election method will be introdced and discssed. It will also be hypothetically applied to the 2013 Bndesta, prodcin a considerable ain in its representativeness. Keywords: olicy representation, representative democracy, direct democracy, elections, coalitions. JEL Classification: D71 1 Introdction: policy representation and irrational votin olicy representation that is, how well the electorate s policy preferences are represented by the party system and the overnment has been stdied for half a centry; for instance, see [Miller and Stokes 1963, itkin 1967, Monroe 1979, Bartels 1991, Hartley and Rssett 1992, Stimson et al 1995, Wlezien 1995, Wlezien 1996, Monroe 1998, Miller et al 1999, Sharpe 1999, Smith 1999, owell 2000, Bde et al 2001, Soroka 2003, Wlezien 2004, Klinemann et al 2006, Cerami 2007, Bde and McDonald 2007, Soroka and Wlezien 2010, Volkens et al 2013]. To monitor policy representation in modern democracies, a special database has been created within the prize-winnin Manifesto project (Berlin), aimed at qantitative content analysis of parties election prorams from more than 50 contries coverin all free, democratic elections since 1945 [WZB 2014]. 1

2 An alternative approach how individal voters match with parties is implemented in a nmber of so-called votin advice applications (VAAs). Followin the Dtch StemWijzer (Vote match) lanched on the internet in 1998, several Eropean contries created sites to help electors compare their policy preferences with party manifestos. They inclde Wahlkabine (Votin booth) in Astria, Glasovoditel (Vote ide) in Blaria, Kohovolit (Whom to vote for) in Czech Repblic, Wahl-O-Mat (Election atomate) in Germany, Vote Match in the K, Te kitvalasztanal? (Whom wold yo choose?) in Hnary, Vote Match in Ireland, Cabine Eletorale (Votin booth) in Italy, rovids (The farsihted) in Latvia, Latarnik Wyborczy (Election lantern) in oland, Testeaza-ti votl! (Test yor vote!) in Romania, Kohovolit (Whom to vote for) in Slovakia, Elecciones.es (Elections.es) in Spain and Smartvote Switzerland [Vote match Erope 2014, Garzia and Marschall 2014]. In 2006, the Free niversity of Amsterdam developed a eneral Election compass whose applications rane from the niversity s works concil to the S president and the Eropean arliament [Kieskompas 2006, E profiler 2009]. This device not only evalates the preference proximity between voters and candidates bt also locates them on a two-dimensional landscape with five-point Left Riht and Conservative roressive axes. Most stdies in policy representation consider one-dimensional models, locatin the policy preferences of parties, overnments and voters either on more-or-less policy axes (in terms of bdetin for defence, edcation, health, etc.) or on the areated left-riht ideoloical axis. However, the modern electorate, parties and overnments locate themselves with respect to nmeros policy isses, makin policy preferences essentially mlti-dimensional and hardly redcible to left riht alinment:... in a nmber of Western societies... no socioeconomic or cltral cleavae is evidently more important and stable than others. To be sre, citizens do not constitte a homoeneos mass that can be divided in any manner by the choices they are offered, bt the social and cltral lines of cleavae are nmeros, crosscttin, and rapidly chanin.... The nmber of floatin voters who do not cast their ballot on the basis of stable party identification is increasin. A rowin sement of the electorate tends to vote accordin to the stakes and isses of each election. [Manin 1997, rinciples of Representative Government, pp. 223, 231] Followin the crrent trends, we attempt to measre policy representation with respect to mltiple policy isses and sest an election procedre that wold enhance it. The first part of the paper, Analysis, is devoted to revealin actal policy preferences of the German electorate and evalatin the crrent representativeness of the political parties, the Bndesta, and its coalitions. For this prpose, we apply the framework from [Tanian 2013a], extendin it to the 2013 Bndesta election and 28 parties instead of five, which makes or analysis more comprehensive and allows s to stdy the case with statistical tools. We compare the positions of the parties, the Bndesta and the coalitions on a nmber of policy isses with the otcomes of pblic opinion polls on the same isses and constrct indices of poplarity (the averae percentae of the poplation represented) and niversality (freqency in representin a majority). The reslts are discorain. The 2013 election winner, the nion of the two conservative parties CD/CS with their 41.6% of the votes, is the least representative amon the 28 parties considered. Of the other three parties eliible for Bndesta seats (with 5% of the votes), only the left party, DIE LINKE, 1 attains the pper-middle rankin, whereas the social democrats, the SD, and the ecoloists, GRÜNE, are located in the low-middle domain. As a reslt the policy representativeness of the 2013 Bndesta is abot 50%, the same as if its position on every isse had been decided by tossin a coin. The coalition formation aravates the sitation frther. 1 The capitalization of the party names follows the way the parties themselves do it. 2

3 This election failre can be explained by the known phenomenon of irrational behavior on the part of the voters [Caplan 2006, Bartels 2008, Bartels 2010], who often act inconsistently with their own policy preferences, disreard party manifestos and clin to political family traditions, even if otdated. A possible explanation is the shift in the German(and world) political spectrm to the riht after the collapse of commnism and the 1990 German renification, while voters contine to believe that the parties represent the same vales as a few decades ao. Finally, it is possible that policy representation is not the primary concern of the electorate. The contradiction with the rational voter model is sometimes interpreted as rational inorance (delvin deeply into the matter for an optimal choice is too costly and makes little sense, especially becase of the neliible role of one vote [Downs 1957, pp , ]), or as expressive votin (since a sinle vote is not decisive, one can vote altristically for its own sake rather than to brin abot particlar conseqences [Brennan and Lomasky 1993, p. 25]). Besides, voters are often inflenced by a priori jdments abot the candidates platforms or pay more attention to the candidates personal imaes than to their actal political profiles and real merits [Adams et al 2011, Manin 1997, pp ]. The seeminly arbitrary behavior of voters is also explained by the existence of isses beyond the political aenda, incldin ideoloical, reliios, ethnic, traditional, cltral and intellectal reasons [Roemer 1998, Groseclose 2001]. Some athors even sppose that electors can intentionally vote strateically for non-favorite candidates [Alesina and Rosenthal 1995, Kedar 2009]. For instance, one who prefers a moderate policy can vote for the left party to shift the conservative stats qo toward the left. This practice of compensatory votin resembles barainin when the opposin sides initially declare extreme startin points in order to end p somewhere in the middle. Another explanation for inconseqent voter behavior can be iven with a reference to the notion of retrospective votin [Bartels 2010, Caplan 2006, p. 107]. This means that if citizens are satisfied with the crrent policy then the overnment is reelected; a disappointed poplation sally votes for the opposition. Indeed, the 2013 reelection of the CD/CS took place aainst the backrond of ood German performance drin the economic crisis, especially compared with other Eropean contries. Toether with the contry s renification nder the overnance of the same CD/CS, the economic sccess contribted to Germany s international inflence and completely restored the political standin rined by World War II, which was hihly appreciated by the poplation. 2 On the other hand, drin the SD administration from , the Hartz labor market reforms fell heavily on the sholders of weak social rops [Hartz concept 2013]. This withdrawal from the traditional solidarity position is the major reason why, from , the SD lost 29% of its members [SD statistics 2013]. Havin been fonded in 1863 as the Allemeiner Detscher Arbeiterverein (the General German Workers Association), the SD had lost mch of its poplar natre, and a 2013 pblic opinion poll revealed that 59% of Germans considered it no loner representative of workers interests [Infratest dimap 2013]. Ths, the 2013 electoral sccess of the CD/CS is also de to the weakenin of its main competitor. Most athors, reconizin the ap between voters preferences and political reality, nevertheless defend the viewpoint that, in a representative democracy, political instittions are well controlled by the pblic. It is ared that the electorate is not as sperficial as it may seem 2 It is however qestionable whether Germany s recent achievements are de to CD/CS overnance. nlike some other indstrialized contries, which trned to services and compter technoloies, Germany did not abandon its top-qality manfactrin niche in the world economy and contined to export its prodcts to contries like China which were relatively naffected by the crisis and had lon-rnnin development prorams for their he internal markets. Ths, Germany s ood performance drin the crisis is rather explained by its traditional adherence to heavy indstries whose prodcts remain in hih demand by emerin economies. Neither shold the renification of Germany be attribted exclsively to the CD/CS; the decisive role has played the SSR headed by Mikhail Gorbachev. 3

4 [Kin 2002, Kedar 2009, Soroka and Wlezien 2010], frthermore relar elections with variable reslts provide adeqate electoral feedback abot the policies prsed [ae and Shapiro 1992, Stimson et al 1995, Manin 1997, owell 2000, Erikson et al 2002, McDonald and Bde 2005, Kedar 2009]. olicymakers are however more concerned than scholars, characterizin the same sitation as a democratic deficit, which is crrently bein discssed at all political levels, incldin the Eropean nion and the nited Nations[Eropean nion 2014, N arliamentary Assembly 2008]; for a srvey see [Mair and Thomassen 2010]. Only recently have scholars started to realize the real problem: We find no sbstantively or statistically sinificant evidence that voters adjst their perceptions of parties Left-Riht positions in response to the policy statements in parties election manifestos a conclsion that is strikin iven that interviews with Eropean political elites that we condcted (discssed below) sest that parties campain on the basis of these manifestos... By contrast, we find that Eropean citizens do react to their perceptions of parties Left-Riht positions, i.e., citizens adjst their Left-Riht positions and their partisan loyalties in response to the parties policy imaes... Bt, becase voter perceptions donot track the parties actal policy statements, there is a disconnect between shifts in elite policy discorse and voter reactions. Voters react to what they perceive the parties stand for, bt these perceptions do not match p with the actal statements in the parties policy manifestos, which form the basis for the parties election campains. [Adams et al 2011, Is anybody listenin? Evidence that voters do not respond to Eropean parties policy statements drin elections, Am J olit Sci: 55, p. 371] The voters sperficiality and inattention to party manifestos wold be not so dramatic if it did not lead to discrepancies between the electorate and the overnment. For instance, the German poplation responded with nprecedented violence to qite ordinary and rational decisions, like constrctin a new railway station in Stttart to replace its termins with modern throh-tracks for hih-speed trains [Alle-een-S21.de 2014], or transportin and disposin nclear waste [Castor-Transport eht af schwieriste Etappe 2010]. The mismatch between the electorate and the actal political administration is clearly revealed by or analysis. A serios warnin to the crrently overnin moderate parties is the sperior policy representation of the increasinly ndifferentiated ranks of the nativist riht and anti-american far-left [Kamm 2016, p. 26]. The simltaneosness of their hih rankin can be explained by the fact that both extreme left and extreme riht parties are poplist (thoh with different backronds), they appeal primarily to the same lower classes, and they exhibit similarities in their positions on many policy isses spported by lare fractions of the poplation. Crrently they arner only few votes, bt if the discrepancy between the citizenry and overnment increases, people can elect extreme politicians. However, to a lare extent election failres o nnoticed, becase politicians, when reelected, receive falty feedback from the poplation and contine to prse npoplar policies. Section 2, Representativeness indices of the parties and the Bndesta, describes the data sorces and defines the indices of poplarity and niversality of German parties and the 2013 Bndesta before coalition formation. It is shown that the winnin parties are not as representative as one miht expect, meanin that voters have not acted consistently with their own policy preferences, and that the Bndesta position is practically nrelated to pblic opinion. In Section 3, Evalation of coalitions, the indices mentioned are extended to possible Bndesta coalitions, and some neative effects of coalition formation on the representativeness of the Bndesta are discssed. 4

5 In Section 4, Conclsions to art I, the main findins of the first part of the paper are recapitlated and placed into context. The Appendix (Section 5) contains fll references to the polls of pblic opinion sed in the stdy, incldin exact formlation of the poll qestions. 2 Representativeness indices of the parties and the Bndesta Table 1 shows the 28 German paries that took part in the 2013 federal election, and Table 2 displays the balance of pblic opinion and party positions on 38 policy isses. We perform or analysis from the standpoint of virtal direct democracy. The positions of the parties are compared with the reslts of pblic opinion polls. Then the parties are evalated with two indices of representativeness poplarity and niversality based on the size of rops resltin from crosscttin cleavaes [Miller 1964, Wriht 1978, Miller 1983, Brams et al 1998]. Both indices sest a kind of correlation measre for estimatin the proximity between party positions and voters opinions [Achen 1977, Achen 1978]. This approach differs from measres of ideoloical conrence at the one-dimensional left riht axis [Golder and Stramski 2010]. The cleavaes considered below are defined in a mlti-dimensional policy space whose dimensions are not directly linked to any ideoloy. The crosscttin cleavaes sed in or analysis are determined by dichotomos qestions, each dividin the society into two rops: protaonists (Yes-rop) and antaonists (No-rop); abstentions are not considered. A party, answerin these qestions, represents the Yes-rops for some qestions and the No-rops for others. By definition, the poplarity index of a party is the averae size of the rop represented, averaed over all the qestions selected. The niversality index of a party is its freqency in representin a majority. The poplarity and niversality indices are compted in two versions: for eqally weihted qestions, meanin their eqal importance, and for the qestions weihted with the base 2 loarithm of thosand Goole hits for the qestions keywords (the keywords in German in order to restrict thesearch to the German area are the titles to the qestions in Appendix) it is assmed that the nmber of relevant search reslts reflects the importance of the qestion. The qestions and the party positions are from the Wahl-O-Mat the votin advice application of the German Federal Aency for Civic Edcation [Bndeszentrale für politische Bildn 2014]. The Wahl-O-Mat (an invented word composed from the German Wahl = election and Atomat) is the German version of the Dtch Internet site StemWijzer ( VoteMatch ) [ro demos 2014], which was oriinally developed in the 1990s to involve yon people in political participation. Both websites help the sers locate themselves on the political landscape by testin how well their opinions match with party positions. Before an election (local, reional, federal and even Eropean), a special overnmental spervisin committee compiles a list of qestions on topical policy isses ( Introdce minimm wae? Yes/No, Introdce a eneral speed limit on motorways? Yes/No, etc.) and asks the parties participatin in the election for their answers. A ser of the site answers the same qestions, eventally attribtin weihts to reflect their importance, and then the proram compares his or her political profile with that of the parties and finds the best-matchin party, the next-best-matchin party, etc. The Wahl-O-Mat provides no statistics for individal answers, bt even if they were available, they wold be biased toward internet sers. 3 For this reason we refer to the pblic opinion polls that pertain to 36 of the 38 Wahl-O-Mat qestions (for two isses, no appropriate polls 3 Before the 2013 Bndesta election, the Wahl-O-Mat site was visited over 13 million times [Bndeszentrale für politische Bildn 2014]. 5

6 Table 1: German parties in the 2013 Bndesta election. Sorce: [Bndeswahlleiter 2014, Bndeszentrale für politische Bildn 2014] arty loo arty description Nmber of members Nmber of votes ercentae of votes nion of Germany s two main conservative parties, Christlich Demokratische nion Detschlands (Christian Democratic nion of Germany) fonded in 1950 and Christlich-Soziale nion in Bayern (Christian Social nion of Bavaria) fonded in 1945 Sozial-demokratische artei Detschlands (Social Democratic arty) fonded in Die LINKE (The Left) fonded in 2007 as the merer of East German commnists and the Electoral Alternative for Labor and Social Jstice (WASG), a left-win breakaway from the SD BÜNDNIS 90/DIE GRÜNEN (Alliance 90/The Greens) fonded in 1993 as the merer of DIE GRÜNEN (West Germany) and BÜNDNIS 90 (East Germany), both with a social-democratic backrond Freie Demokratische artei (Free Democratic arty) fonded in 1948, liberal political party close to employers oranizations Alternative für Detschland (Alternative for Germany) fonded in 2013, a conservative, ero-crrency-sceptic party iratenpartei Detschland (irate arty of Germany) fonded in 2006, part of the international irate movement promotin the information society with free access to all diital media National-demokratische artei Detschlands (National Democratic arty of Germany) fonded in 1964, a far-riht German nationalist party FREIE WÄHLER (Free Voters) fonded in 2009, a party in opposition to the E financial policy Mensch mwelt Tierschtz (Hman Environment Animal Welfare) fonded in 1993, a party promotin the introdction of animal rihts into the German constittion Ökoloisch-Demokratische artei (Ecoloical Democratic arty) fonded in 1982, a conservative environmentalist party Die Repblikaner (The Repblicans) fonded in 1983, a national conservative party opposin to immiration artei für Arbeit, Rechtstaat, Tierschtz, Elitefördern nd basisdemokratische Initiative (arty for Work, Rle-of-Law, rotection of Animals, Advancement of Elites and Grass-root Democratic Initiative) fonded in 2004, a poplist parodical party with totalitarian trends 6

7 Table 1: (contined) German parties in the 2013 Bndesta election arty loo arty description Nmber of members Nmber of votes ercentae of votes Bürerbewen pro Detschland(ro-Germany Citizens Movement) fonded in 2005, a far-riht poplist party opposin illeal immiration and mlti-national corporations and financial instittions Bayernpartei (Bavaria arty) fonded in 1946, a separatist Bavarian party advocatin Bavarian independence within the Eropean nion Volksabstimmn (Referendm party) fonded in 1997, a party promotin direct democracy of the Swiss type RENTNER artei Detschland (German arty of ensioners) fonded in 2002, a party of social welfare state bridin the interests of enerations artei der Vernnft (arty of Reason) fonded in 2009, a liberal party promotin the ideas of the Astrian School of economic 1000 thoht minimal state, free market, decentralization of political power and sbsidiarity Marxistisch-Leninistische artei Detschlands (Marxist-Leninist arty of Germany) fonded in 1982, an anti-revisionist party, referrin to Marx, Enels, Lenin, Stalin and Mao Zedon artei Bibeltreer Christen (arty of Bible-abidin Christians) fonded in 1989, a conservative evanelical party, opposin anti Semitism, same-sex marriae and abortion Bündnis für Innovation nd Gerechtikeit (Alliance for Innovation and Jstice) fonded in 2010, a party of Mslims promotin their interation Bürerrechtsbewen Solidarität (Civil Rihts Movement Solidarity) fonded in 1992, a part of the worldwide LaRoche (.S. politician) Yoth movement with repblican orientation bt promotin worlwide solidarity, e.. abolishin the debts of the Third World DIE FRAEN (The Women), a feminist party fonded in promotin the rihts of women artei der Nichtwähler (arty of Non-Voters) fonded in 1998, a party with a social democratic backrond promotin improvin representative democracy by introdcin elements of direct democracy Bündnis 21 / Rentnerinnen- nd Rentner-artei (Alliance 21 / arty of Retirees) fonded in 2007, promotin improvin the pension, health and edcation systems Die Violetten für spiritelle olitik (The Violets for spirital olicy) fonded in 2001, claimin to represent alternative spirital politics in the new ae Familien-artei Detschlands (The Family arty of Germany) 600 fonded in 1983, a party promotin family vales artei für Soziale Gleichheit, Sektion der Vierten Internationale (arty of Social Jstice, Section of the Forth International) fonded in 1997, a Trotskyist party 7

8 Table 2: blic opinion and party positions on topical policy qestions. Sorces: [Bndeszentrale für politische Bildn 2014] and others, see Appendix. Qestions Qestion weihts blic opinion n- Goole ro- Anweiht-hited tao- tao- in K nists nists 1 lo 2 % % 1 Introdce a nationwide minimm wae Childcare sbsidy for parents whose kids do not attend state-sponsored day care 3 Introdce a eneral speed limit on hihways Germany shold retain the Ero as its crrency Electricity prices shold be more heavily related by the state Video srveillance in pblic spaces shold be expanded Germany shold introdce an nconditional basic income Only oranic aricltre shold receive financial incentives All children, reardless of cltral heritae, shold receive eqal edcation The top income tax rate shold be increased Germany shold leave NATO No new constrction of coal-fired enery plants The mornin after pill mst be available by prescription only All banks in Germany shold be nationalized Germany shold accept more refees State compensation for the time employees spend carin for incapacitated rel-atives 6.58?? 17 olitical parties that are nconstittional shold contine to be illeal The level of federal stdent financial aid shold be independent of the parents income 19 Border control shold be re-introdced Institte a leal qota for women on company overnin boards Financially stroner federal states shold less spport weaker federal states The leally mandated retirement ae shold be lowered aain The overnment shold employ more people with immirant backronds ?? 24 Exports of mnitions shold be forbidden Retain the tax law that favors the spose who earns mch less than the other spose or nothin 26 Germany shold champion Trkey s bid for E membership Bndesta members shold reveal their spplementary incomes to the last Ero 28 Enery-intensive indstries shold bear more of the transition to renewable enery costs 29 Redce lon-term nemployment benefits for those who trn down a job offer The state shold contine to collect tithes on behalf of reliios instittions Reqire all citizens to enroll in the pblic health insrance system Every state in the Ero zone shold be liable to pay its own debts Homosexal coples shold be allowed to adopt No collection of commnication data (e.. telephone, internet) withot prob-able case 35 Limit rent price increases, also pon trnover of renters German citizens shold not be allowed to hold additional nationalities Institte a passener-car toll on the national hihways Introdce referenda at the federal level

9 Table 2: (contined) blic opinion and party positions on topical policy qestions Qestions arty positions: 1 Yes; 0 No;? Missin data CD/CS SD DIE LINKE GRÜNE FD AfD IRATEN ND FREIE WÄHLER Tierschtzpartei ÖD RE Die ARTEI pro Detschland B Volksabstimmn RENTNER artei der Vernnft MLD BC BIG BüSo DIE FRAEN Nichtwähler Bündnis 21/RR DIE VIOLETTEN FAMILIE SG ? ? ? ? ? ? ? 1 1 0? ? 0 0? ? ? ? ?? ?? ? 0 1 0?? ??? ? ? 1 0 1? 1 1 1? ? ? ? 0 1 0? ? ? ? 0 1? 0 0 1? 1 0 1? 1 1 1? ? ? ? ? ? ?? 0 1? 1 1 1?? ? ? 1 1 1? ? ? ? ? 0 20? ? ? 0?? 1? ? ??? 0? 1 0? 1? 0 1 0? 0 0 0? ? 0? ?? 1 1? ? ?? ?? ? ? 1 1?? 0 0 1? 0? ? ? 1 0? ? 0? 0 0 0? ??? ? ? ? 1? ? ?? 1 0? ? ? 0? ? ? ? ? ? ? 1? ? ? ? ? ? ? 0 0? 0? ? ??

10 were fond) and which were condcted between the two most recent federal elections in 2009 and A few poll qestions are not formlated exactly as in the Wahl-O-Mat: No. 11 on Germany s leavin NATO is simlated by the poll qestion on the end of NATO s defence service and the necessity of its fndamental reoranization; No. 14 on nationalization of banks, is simlated by the poll qestion on the nationalization of banks in the case of emerency; No. 32 on the liability of every Ero-zone state to pay its own debts is simlated by a similar poll qestion that explicitly mentions Spain and Romania; and No. 34 on preventive collection of commnication data is simlated by the poll qestion abot preventive commnication data collection by secret services of the SA and Great Britain. The oldest (2009) polls are: No. 5 on the relation of electricity prices by the state; No. 9 on separate schools; No. 14 on nationalization of banks; and No. 29 abot sanctions aainst the nemployed who draw benefits bt are not actively seekin work. Three polls are internet srveys that are less representative than those condcted by reconized statistical offices: No. 18 on state sbsidies for stdents; No. 19 on restorin the border control; and No. 21 abot financial eqalization of states. For the exact formlation of the qestions in the pblic opinion polls, their dates and otcomes see the Appendix. Of corse, the conclsions based on data with deficiencies reqire reservations. On the other hand, de to a certain interdependence between responses to different qestions, or data collection is somewhat redndant; hence, the conclsions can be expected to be relatively insensitive to data inaccracies. In any case, or stdy is aimed rather at exposin a problem than formlatin final jdements. Let s illstrate the constrction of representativeness indices with Fire 1. It displays the 38 Wahl-O-Mat qestions toether with their oriinal nmbers and color bars indicatin the balance of pblic opinion and the positions of the for Bndesta parties (that received 5% of the electoral votes); if the data were missin, the correspondin rectanles are missin as well. For a specific example, we take the first qestion, Introdce a nationwide minimm wae. The balance of pblic opinion, with 86% for (protaonists) and 12% aainst (antaonists) is shown by the ble bar. Its lenth is normalized, that is, the abstentions have been removed from consideration, and the total bar lenth is broht p to 100%. The bar s sements to the left and to the riht of the central axis are proportional to the percentae of antaonists and protaonists in the society, respectively. Ths the prevailin pblic opinion is visalized by the ble bar s bias from the center. Every Bndesta faction is depicted by a rectanle in the official party color; its lenth is proportional to the nmber of seats it holds in the Bndesta. The No/Yes party opinion on the qestion is reflected by the positionin of the rectanle to the left or to the riht of the central axis, respectively. For every qestion, the iven party represents a certain percentae of the poplation, as revealed by the correspondin pblic opinion poll. For instance, the nion CD/CS with its No answer to Qestion 1 represents the opinion of 12% of the poplation verss the 86% who answered Yes. After removal of abstentions and normalization (brinin the total to 100%), we obtain the CD/CS representativeness for Qestion 1: r CD/CS,1 = Similarly, with the Yes answer to Qestion 2, Childcare sbsidy for parents whose kids do not attend state-sponsored day care, the nion CD/CS expresses the opinion of 20% of the poplation verss 77%. After removal of abstentions and normalization we obtain the followin CD/CS representativeness for Qestion 2: r CD/CS,2 = , and so on. Takin the averae representativeness of the CD/CS over the 32 qestions for 10

11 CD/CS SD DIE LINKE GRÜNE blic opinion 1 Introdce a nationwide minimm wae 2 Childcare sbsidy for parents whose kids do not attend state-sponsored day care 3 Introdce a eneral speed limit on hihways 4 Germany shold retain the Ero as its crrency 5 Electricity prices shold be more heavily related by the state 6 Video srveillance in pblic spaces shold be expanded 7 Germany shold introdce an nconditional basic income 8 Only oranic aricltre shold receive financial incentives 9 All children, reardless of cltral heritae, shold receive eqal edcation 10 The top income tax rate shold be increased 11 Germany shold leave NATO 12 No new constrction of coal-fired enery plants 13 The mornin after' pill mst be available by prescription only 14 All banks in Germany shold be nationalized 15 Germany shold accept more refees 16 State compensation for the time employees spend carin for incapacitated relatives 17 olitical parties that are nconstittional shold contine to be illeal 18 The level of federal stdent financial aid shold be independent of the parents' income 19 Border control shold be re-introdced 20 Institte a leal qota for women on company overnin boards 21 Financially stroner federal states shold less spport weaker federal states 22 The leally mandated retirement ae shold be lowered aain 23 The overnment shold employ more people with immirant backronds 24 Exports of mnitions shold be forbidden 25 Retain the tax law that favors the spose who earns mch less than the other spose or nothin 26 Germany shold champion Trkey's bid for E membership 27 Bndesta members shold reveal their spplementary incomes to the last Ero 28 Enery-intensive indstries shold bear more of the transition to renewable enery costs 29 Redce lon-term nemployment benefits for those who trn down a job offer 30 The state shold contine to collect tithes on behalf of reliios instittions 31 Reqire all citizens to enroll in the pblic health insrance system 32 Every state in the Ero zone shold be liable to pay its own debts 33 Homosexal coples shold be allowed to adopt 34 No collection of commnication data (e.. telephone, internet) withot probable case 35 Limit rent price increases, also pon trnover of renters 36 German citizens shold not be allowed to hold additional nationalities 37 Institte a passener-car toll on the national hihways 38 Introdce referenda at the federal level NO YES -100% -50% 0 50% 100% ercentae of NO/YES votes (abstensions have been omitted) Fire 1: blic opinion and representation thereof by the 2013 Bndesta 11

12 which we have both opinion poll data and definitive party answers, we obtain the party s nweihted poplarity index CD/CS = 32 q=1 r CD/CS,q 32 = ( 100% 40%). 32 Replacin the weihts 1 32 in the above formla with the normalized Goole-weihts (that is, proportionally redced to the total = 1), we obtain the Goole-weihted version of poplarity. It shold be noted that for each party we consider only the qestions backed p with pblic opinion polls and definitive party answers, so the normalization of weihts varies from one party to another. The CD/CS (nweihted) niversality index is the percentae of the qestions for which the CD/CS represent a non-strict majority ( 50%). Aain, we restrict orselves to the 32 qestions with known CD/CS positions and backed p by pblic opinion polls. Since the CD/CS represent a non-strict majority on 11 ot of 32 qestions, CD/CS = q:r CD/CS,q = q=1 ] rond [r CD/CS,q 32 = ( 100% 34%). If the qestions are conted with Goole-weihts instead of 1/32, we obtain the Goole-weihted version of the niversality index. The indices of other parties are compted in the same way, bein based on the party positions in Table 2. The indices of the Bndesta before coalition formation are based on the Bndesta majority opinions on the isses. For instance, the Bndesta s position on Qestion 1 is the one of the majority of three YES-factions, of SD, DIE LINKE and GRÜNE, as visalized by the cmlative lenth of the party bars in Fire 1 that srpasses the 50%-threshold shown by dotted lines. Fire 2 displays the indices of poplarity and niversality for 28 German parties and the Bndesta, each in two versions: for nweihted qestions (marked with ) and Gooleweihted qestions (marked with ). The parties are sorted in decreasin order of the mean of the for indices. The mean index makes sense, becase the poplarity and niversality indices are correlated, bein defined as expected representativeness and expected ronded representativeness, respectively. Table 3 shows that all the rank correlations between the representativeness indices are very close to In particlar, the correlation of 0.99 between the nweihted and the Goole-weihted indices means that the party ranks are not very sensitive to the qestion weihtin. This can be explained as follows. The responses of a party are backed p by the party s ideoloy, which predetermines most of the party answers. The same applies to the electorate rops. With reservations, a party represents more or less the same electorate rop on most of the qestions. One can even add or remove some qestions withot reatly affectin the indices. All of these make the role of qestion weihtin in party indexin rather small which is exactly what we observe. De to the hih correlation of the indices we can characterize the parties by their mean index, as in Fire 2, or by the most impartial, nweihted index. Table 3 also shows that the correlation between the parties indices of representativeness and the votes received is neative, thoh not statistically hihly sinificant. This means that the electorate pays little attention to policy representation, often votin contrary to its own policy preferences. The election winner, the nion CD/CS, has the lowest policy representation rank (28 ot of 28) bt received the most votes, 41.5%. Moreover, the 34 40% rane of its 12

13 oplarity () and niversality (), in % oplarity () and niversality (), in % oplarity () and niversality (), in % Nichtwähler Rank: 1 Votes: 0.03% 74 ÖD Rank: 15 Votes: 0.29% Volksabstimmn Rank: 3 Votes: 0.07% RENTNER Rank: 16 Votes: 0.06% MLD Rank: 12 Votes: 0.06% GRÜNE Rank: 17 Votes: 8.45% BIG Rank: 18 Votes: 0.04% ND Rank: 6 Votes: 1.28% SD Rank: 19 Votes: 25.73% DIE VIOLETTEN Rank: 13 Votes: 0.02% pro Detschland Rank: 14 Votes: 0.17% FREIE WÄHLER Rank: 20 Votes: 0.97% BC Rank: 21 Votes: 0.04% Fire 2: Sheet A. Indices of German parties and the Bndesta: poplarity, niversality, for nweihted qestions, and for qestions weihted by the nmber of Goole hits 13 DIE LINKE Rank: 7 Votes: 8.59% IRATEN Rank: 11 Votes: 2.19% Die ARTEI Rank: 5 Votes: 0.18% DIE FRAEN Rank: 4 Votes: 0.03% Tierschtzpartei Rank: 10 Votes: 0.32% SG Rank: 9 Votes: 0.01% FAMILIE Rank: 8 Votes: 0.02% Bündnis 21/RR Rank: 2 Votes: 0.02%

14 oplarity () and niversality (), in % B Rank: 22 Votes: 0.13% BüSo Rank: 23 Votes: 0.03% FD Rank: 24 Votes: 4.77% RE Rank: 25 Votes: 0.21% artei der Vernnft Rank: 26 Votes: 0.06% AfD Rank: 27 Votes: 4.70% CD/CS Rank: 28 Votes: 41.55% BNDESTAG Mean index: 50 Votes: 84.32% Fire 2: (contined) Indices of German parties and the Bndesta: poplarity, niversality, for nweihted qestions, and for qestions weihted by the nmber of Goole hits Table 3: earson rank correlation between parties votes and their indices of representativeness Votes nweihted Goole-weihted nweihted Goole-weihted Mean index Votes nweihted Goole-weihted nweihted Goole-weihted Mean index vale <-vale <-vale 0.10 indices says that the CD/CS is more non-representative than representative. At the same time, the most representative amon the parties eliible for Bndesta seats, DIE LINKE, whose indices attain 64 77%, received only 8.6% of the votes; see Fire 2. The Bndesta s policy representation is reatly inflenced by the low representativeness of the CD/CS, makin the Bndesta s indices before coalition formation as low as ca. 50%; see Fire 2. It shold be noted that the same index vales wold be expected if the matchin of the Bndesta s position with the pblic opinion on every isse had been decided by tossin a coin. All of these mean that the 2013 Bndesta s policy profile is practically independent of pblic opinion. 3 Evalation of coalitions Table 4, illstrated by Fire 3, characterizes the minimal eliible coalitions, that is, those possible to form by the parties with 5% of the votes and constittin a Bndesta majority. In real politics, the eliible coalitions are sally minimal, i.e., they contain no more parties 14

15 Table 4: Indices of minimal eliible coalitions compted for nweihted qestions and the impact of member weihts on the coalition decisions p = 0.50 Coalitions Bndesta nanimity oplarity niversality seats Expectation Standard deviation Expectation Standard deviation %/Rank %/Rank %/Rank %/Rank %/Rank %/Rank 1 CD/CS SD 79.8 / / / 2 ±2.9 / / 2 ±5.8 / 3 2 CD/CS DIE LINKE 59.5 / / / 4 ±2.0 / / 4 ±4.2 / 1 3 CD/CS GRÜNE 59.3 / / / 3 ±2.5 / / 3 ±4.9 / 2 4 SD DIE LINKE GRÜNE 50.7 / / / 1 ±3.4 / / 1 ±6.4 / 4 CD/CS SD DIE LINKE GRÜNE 80 4 nanimity, in % oplarity, in % niversality, in % Fire 3: Indices of minimal eliible coalitions compted for nweihted qestions and the impact of member weihts on the coalition decisions p = The lenths of the colored bars are proportional to the faction sizes 15

16 than necessary, becase the more parties, the more complex the neotiations and the less power enjoyed by each faction (cf. with Riker s minimm winnin coalitions [Riker 1962]). For instance, the coalition CD/CS SD GRÜNE is eliible bt not minimal; that is, the coalition CD/CS SD, bein itself eliible, does not need GRÜNE. The first colmn of Table 4 enmerates the minimal eliible coalitions shown in Fire 3 by bars in theofficial party colors wherethelenth is proportional to thenmberof votes received by the party. The second colmn shows the coalition weiht in % of parliament seats. The percentae of the parliament seats is proportional to the votes the parties received in the election (small adjstments to the nmber of parliament seats as prescribed by the German constittion are not taken into accont). Therefore, the coalition SD DIE LINKE GRÜNE, with 42.8% of the electors votes, has 50.7% of the parliament seats, constittin a Bndesta majority: Weiht of SD LINKE GRÜNE in the Bndesta = ( 100% 50.7%). The third colmn of Table 4 contains the deree of nanimity of the coalition expressed in the total weiht of the qestions on which all the coalition members aree. In Fire 3, the coalitions nanimity corresponds to the heihts of the flastaffs. As shown previosly, qestion weihtin plays a secondary role, so we refer to the least partial eqal weihts of the qestions; correspondinly, the coalition indices of poplarity and niversality are also nweihted; they are X Y coordinates of the flastaffs. Ifacoalition is nanimoson a certain qestion then its answer is thesameas of every member. If coalition members disaree on an isse then the probabilities of the coalition s Yes/No answer to this qestion cold be assmed proportional to the protaonist-to-antaonist ratio within the coalition. As expressed in a personal conversation with the athor by Tobias Lindner, Bndesta member (GRÜNE), the reality is even more ncertain. To deal with the ncertainty in coalition decision makin, we introdce the parameter p proportionality of inflence to size of the faction 0 p 1, which we explain here with an example. Sppose that, for a certain qestion, the protaonist-to-antaonist ratio within a coalition is 3:1, that is, the Yes-faction is three times larer than the No-faction. The p = 1 denotes the exact proportionality of inflence to size, when the coalition answers Yes with the probability eqal to the weiht of the Yes-faction 3/(3+1) = 3/4, and No with the probability eqal to the weiht of the No-faction 1/4. The p = 0 denotes no proportionality of inflence to size, that is, the coalition adopts each alternative opinion with eqal chances 1/2. The medim ncertainty p = 1/2 means that the inflence of faction sizes on the coalition answer is a mix of the two extreme cases in proportion p = 1/2 and 1 p = 1/2: rob (Yes) = p }{{} 1/2 rob (No) = p }{{} 1/2 3/4 }{{} Yes-faction s size inflence 1/4 }{{} No-faction s size inflence +(1 p) }{{} 1/2 +(1 p) }{{} 1/2 1/2 = 5/8 }{{} No inflence of size 1/2 = 3/8. }{{} No inflence of size Throhot this paper, the medim ncertainty p = 1/2 is assmed and applied to all coalitions considered frther. nder ncertainty, the coalitions indices of poplarity and niversality trn into random variables. Then the coalition s poplarity and niversality are nderstood as the expected size of the 16

17 rop represented, and as the expected freqency in representin a majority, respectively. These indices are no loner exact manitdes bt expected estimates, with their standard deviation rearded as the estimation accracy. The coalition indices and their standard deviations are compted with formlas from [Tanian 2014, p. 338]. Note that in Table 4, the hiher ranks are assined to reater representation indices, bt to smaller standard deviations characterizin their hiher accracy. A hih deree of nanimity facilitates coalition formation, becase parties with close positions cooperate more easily. However, the most nanimos and simltaneosly most representative minimal eliible coalition, SD LINKE GRÜNE, failed becase of tensions between the SD and DIE LINKE arisin from some of DIE LINKE s politicians who left the SD in Amon other minimal eliible coalitions (constitted by CD/CS with the SD, DIE LINKE, or GRÜNE) the coalition CD/CS SD has the hihest nanimity and, at the same time, the hihest representativeness. This is the coalition that is crrently overnin. 4 Conclsions to art I Inconsistency of election reslts with the voters policy preferences. The neative correlation between the parties policy representation indices and the votes received means that German citizens vote rather inconsistently with their policy preferences. The Bndesta s disconnection with pblic opinion. The policy representation indices of the 2013 Bndesta before the coalition formation are all abot 50%. This means that the Bndesta s positions on policy isses are statistically independent of pblic opinion. Frther deradation of representativeness throh coalition bildin. The overnin coalition, CD/CS SD, has policy representation indices of ca. 45%, lower than the Bndesta before the coalition formation. Therefore, coalition bildin can redce the parliament s policy representation. olitical warnin. When re-elected, the moderate overnin parties et falty feedback on their npoplar policies. If the discrepancy between the poplation and the overnment increases, it sets the stae for more representative bt also more extreme parties to et elected. Neliible role of qestion weihtin. The policy representation indices are relatively insensitive to qestion weihtin. Therefore, the mean index or impartial nweihted index can be sed. 5 Appendix. blic opinion polls relevant to the Wahl-O-Mat qestions References [1] Minimm wae. Introdce a nationwide minimm wae. Srvey qestion: Einführn eines esetzlichen Mindestlohns Befürwortn oder Ablehnn? (Infratest dimap, 2013, Jn 3 4, 1003 interviews; pros 86%, cons 17

18 12%). mindestlohn-raf.pdf. Cited 3 Oct 2013 [2] Childcare sbsidy. Childcare sbsidy for parents whose kids do not attend state-sponsored day care Srvey qestion: Betrenseld Ach künfti zahlen oder für Asba von Kitas verwenden? (Infratest dimap, 2013, Jl, 1003 interviews; pros 20%, cons 77%). infratest-dimap.de/fileadmin/ mirated/content ploads/dt1308 bericht 01.pdf. Cited 11 Apr 2016 [3] Speed limit. Introdce a eneral speed limit on hihways Srvey qestion: Einführn eines enerellen Tempolimits von 120 oder 130 km/h af den detschen Atobahnen (Infratest dimap, 2012, 9 10 Oct, 1000 interviews; pros 53%, cons 49%). mfraen/aktell/ard-markencheck-adac/. Cited 4 Oct 2013 [4] Ero. Germany shold retain the Ero as its crrency Citation: Wie eine repräsentative mfrae des Forsa-Institts für das Handelsblatt (Dienstaasabe) zeit, sind hete 69 rozent der Detschen dafür, das nee Geld z behalten. Nr 27 rozent wollen wieder zrück zr D-Mark. politik/detschland/exklsivmfrae-detsche-finden-den-ero-t/ html. Cited 3 Oct 2013 [5] Electricity prices. Electricity prices shold be more heavily related by the state Srvey qestion: Halten Sie die reise für Enerie- nd Heizkosten für fair? (2009, Nov, 1225 interviews; pros 90%, cons 10%) /mfrae/bewertn-der-preise-fer-enerie-nd-heizn/. Cited 3 Oct 2013 [6] Video srveillance. Video srveillance in pblic spaces shold be expanded Srvey qestion: Asweitn der Videoüberwachn af öffentlichen lätzen nd in Bahnhöfen (Infratest dimap 2012, Dec, 1000 interviews; pros 81%, cons 18%). rosse-mehrheit-der-berer-fer-verstaerkte-videoeberwachn/. Cited 4 Oct 2013 [7] Basic income. Germany shold introdce an nconditional basic income Srvey qestion: Was halten Sie von einem bedinnslosen Grndeinkommen für alle Bürer? (Minle, 9 13 Feb 2012, 1037 interviews; pros 80%, cons 20%) /80-der-detschen-wollen-dasbedinnslose-rndeinkommen/. Cited 4 Oct 2013 [8] Oranic aricltre. Only oranic aricltre shold receive financial incentives Srvey qestion 3: What do yo think is most important that tax-payers money paid to farmers shold be sed for? [sic] (TNS-Germany, Jan 2013, 1003 interviews; pros 76%, cons 23%). opinion on the se of pblic money in the aricltral sector.pdf. Cited 04 Oct 2013 [9] Separate school edcation. All children, reardless of cltral heritae, shold receive commnal edcation Srvey qestion: Finden Sie die Fordern nach Einführn einer Einheitsschle richti? (Forsa, Nov 2009, 1013 interviews; pros 33%, cons 60%). 18