UC Irvine CSD Working Papers

Transcription

1 U Irvne SD Workng Papers Ttle The Tyranny of the Super-Majorty: How Majorty Rule Protects Mnortes Permalnk Author McGann, Anthony J. Publcaton Date escholarshp.org Powered by the alforna Dgtal brary Unversty of alforna

2 SD enter for the Study of Democracy An Organzed Research Unt Unversty of alforna, Irvne Ths paper demonstrates that majorty rule s the decson rule that provdes most protecton for the worst-off mnorty. Dahl 1956, 1988 argues that the values of popular soveregnty and poltcal eualty dctate the use of majorty rule. However, there are other values that we need to take nto account besdes popular soveregnty and poltcal eualty, notably the protecton of mnorty rghts and stablty. Thus t s commonly argued that there s a tradeoff between poltcal eualty maxmzed by majorty rule and mnorty protecton better provded by systems wth external checks and balances, whch reure more than a smple majorty to enact legslaton. Ths paper argues that ths trade-off does not exst and that actually majorty rule provdes most protecton to mnortes. Furthermore t does so precsely because of the nstablty nherent n majorty rule. Majorty rule s the only decson rule that completely satsfes poltcal eualty. May 1952 shows that majorty rule s the only postvely responsve votng rule that satsfes anonymty all voters are treated eually and neutralty all alternatves are treated eually. If we use a system other than majorty rule, then we lose ether anonymty or neutralty. That s to say, ether some voters must be prvleged over others, or some alternatve must be prvleged over others. Wth super-majorty votng, the status uo s prvleged f there s no alternatve for whch a super-majorty votes, the status uo s mantaned. Followng Rae s 1975 argument, gven that the status uo s more desrable to some voters than to others, some voters are effectvely prvleged. It s certanly the case that super-majorty rules can prvlege protect, f you prefer some voters. Unfortunately, t s not possble to prvlege every group over every other group. If super-majorty rules create a prvleged group, there must be a correspondng under-prvleged group. Nevertheless, super-majortaran decson rules are wdespread, both explctly and mplctly. For example, n the Unted States explctly super-majortaran rules exst n the form of the 60% cloture reurement to end a flbuster n the Senate, the two-thrds reurement to overrde a presdental veto and n the need for a supermajorty to amend the onsttuton. Implctly, the exstence of two legslatve chambers wth dfferent bases of representaton s super-majortaran, n that more than 50% of the popular support s lkely to be needed to ensure a majorty n both chambers. The ommttee system has a smlar effect, to the extent that commttees are able to act as gatekeepers, able to hold-up consderaton of legslaton. Outcomes decded by judcal revew also rest on a super-majortaran bass, n that onsttutonal amendments reure a super-majorty. The number of democraces wth smple majorty rule legslatures wth few external checks s actually ute small, lmted mostly to the small countres of Europe. Super-majortaran decson makng rules have been justfed n terms of the need to protect mnortes from the tyranny of the majorty. In the Unted States, ths argument s

3 1 assocated wth James Madson perhaps napproprately, I wll argue and John. alhoun. Buchanan and Tullock 1962 provde a formalzaton of ths lne of thought, argung that the unanmty rule maxmzes the protecton of ndvdual rghts and economc effcency, and that super-majortaran rules are a second-best approxmaton to unanmty. Ths paper shows formally that as we move from majorty rule towards unanmty, the ablty of mnortes to defend themselves by overturnng unfavorable outcomes s dmnshed, and that therefore majorty rule offers most protecton. Before proceedng two clarfcatons are needed. Frstly, majorty rule should not be euated wth sngle-member dstrct pluralty frst-past-the post electoral systems. Majorty rule s a rule for choosng between two alternatves, where the alternatve wth most votes s chosen. In representatve democraces, majorty rule s used to decde between alternatves before the legslature, usually usng an amendment procedure. There s no reason why those legslators cannot be elected by proportonal representaton PR. Indeed f the reason for adoptng majorty rule n a legslature s to maxmze poltcal eualty, proportonal representaton s the natural system to elect the members of that legslature. As van der Hout 2002 shows, f an electoral system satsfes anonymty, neutralty and some techncal reurements, t must produce results smlar to lst PR. It s unfortunate that the term majortaran s n wdespread use followng jphart 1984 to refer to the wnner-takeall arrangements typcal of the Brtsh Westmnster system of government. It s ronc that the some of the countres that most closely approxmate majorty rule n ther legslatures are those that jphart characterzes as consensual democraces, such as the Netherlands, Sweden, Denmark and Norway. Secondly, majorty rule does not mean that there s necessarly one partcular majorty that rules. One of the central fndng of socal choce theory s that majorty rule s unstable and prone to cycles that s, stuatons where alternatve a beats b, b beats c, but c beats a. Smlarly majorty rule s prone to cyclng coaltons. Because of ths, t s often unwse under majorty rule to try to rule wth a mnmal wnnng coalton and exclude everyone else. If a group s excluded from the wnnng coalton, they can attempt to splt the wnnng coalton by offerng some of ts members more favorable terms than they are currently recevng. Under majorty rule t s ute possble ndeed lkely to see broad coaltons and norms of nclusve, consensual decson-makng. For example, Wengast 1979 shows that a norm of unversalsm, as opposed to mnmal wnnng coaltons, s the optmal soluton to the majorty rule game facng members of the US House of Representatves. In a smlar ven, Dahl 1956 argues that n a polyarchy, we wll not see the tyranny of the majorty, because a sngle, permanent, cohesve majorty wll not exst. The frst secton of ths paper revews the lterature on super-majortaran decson makng. The second llustrates how super-majortaransm can produce perverse results. The thrd secton formally analyzes the logc of super-majortaran decson-makng. The fourth presents the proof that majorty rule offers the greatest protecton to mnortes. Secton 5 consders the effect of uncertanty about the future on these consderatons, whle Secton 6 deals wth the nstablty of majorty rule resultng from cyclng, and why ths s essental to the protecton of mnortes. Super-Majorty Rule And Democracy The theory that checks and balances are needed to restran majorty rule thus producng a system that s effectvely super-majortaran s freuently ascrbed to James Madson.

4 2 However, as Rae 1975 argues, ascrbng such a vew to Madson s problematc, gven that Madson sought a strong natonal government, capable of decsve acton and able to overcome the mmoblsm of the Artcles of onfederaton. 1 As we are all famlar wth Federalst 10, Madson dentfes the problems of mnorty and majorty tyranny. The republcan prncple.e., majorty rule protects aganst mnorty tyranny. However, the only soluton to majorty tyranny gven n Federalst 10 s to have a large extended republc where a sngle cohesve majorty would not exst, a soluton completely compatble wth majorty rule. 2 It s not untl Federalst 51 that Madson advocates external checks and balances, n the context of a drectly elected Presdent. It s notable that Madson dd not support a drectly elected Presdent at the onsttutonal onventon untl md-july 1787, mmedately after he had lost the argument about eual representaton for the States n the Senate. Furthermore, Madson clearly opposed the prncple of super-majortaransm argung that t reversed the prncple of free government and eualed mnorty rule 3, as well as many of the consttutonal features we now consder checks and balances. 4 The orgnal Vrgna plan presented to the onventon, authored prmarly by Madson, was essentally a majorty rule parlamentary system wth the executve chosen by the legslature. Madson gves us, at best, an ambguous justfcaton for restranng majorty rule wth checks and balances; accordng to Rae 1975, t s John. alhoun 1842/1982, 1850/1943 who gves us an uneuvocal theory. Socety s made up a varous classes of people, any of whch may wsh to ntrude on the rghts of others. A system of concurrent majortes, whereby the approval of a majorty of each class s reured for acton, can prevent ths from happenng. alhoun argues that varous features of the US onsttuton most notably eual representaton for the States n the Senate embody ths prncple. Indeed more recent scholars, such as Wengast 1989 and Aldrch 1995 have analyzed the way n whch nsttutons such as North-South party n the Senate and the norm of tcket balancng by the partes essentally provded the southern States wth a veto untl the 1850s. Dahl 1956 s crtcal of what he terms the Madsonan theory of democracy essentally the vew that checks and balances are reured to restran majorty rule. Madson reles on majorty rule to protect aganst mnorty tyranny. Dahl argues that n cases where postve government acton s reured to protect rghts, restranng majorty rule wth checks and balances undermnes ths protecton. Ironcally ths echoes the argument made n the secton from Federalst 58 cted footnote 3. Furthermore, Dahl argues that there s no emprcal evdence that rghts are better protected by the Amercan poltcal system than by European consttutons wth far fewer consttutonal checks, and that nsttutons such as the flbuster, eual representaton n the Senate and judcal revew have been used far more freuently to frustrate the extenson of fundamental rghts than to protect them, most notably n the case of cvl rghts n the South. Buchanan and Tullock 1962 provde a dfferent justfcaton for supermajortaransm. Gven a predetermned allocaton of rghts and property, 5 the decson makng rule that best protects ths allocaton s naturally unanmty. Furthermore, unanmty s the only rule that guarantees that the outcome wll be economcally effcent n the sense of beng Pareto-optmal t s not possble to make anyone better off wthout makng someone else worse off. If unanmty s mpossble because of decson-makng costs, supermajortaran rules may be the second best soluton, n that they provde more protecton than majorty rule aganst costs mposed by socety on ndvduals.

5 3 Rae 1975 crtues Buchanan and Tullock on several grounds. Unanmty only mnmzes the costs socety mposes on ndvduals f we make the strong assumpton that an unwanted polcy mposes a far greater cost on ndvduals than not gettng a polcy that s needed. Rae 1969 shows that f we assume these costs are eual, majorty rule s optmal. Furthermore, Rae crtczes the concept of Pareto optmalty as essentally lockng n the status uo and beng blnd to dstrbutonal consderatons. 6 Most sgnfcantly, Rae shows that unversal consent s logcally mpossble when a decson even f t s to take no acton has to be taken. If there s dsagreement and a decson has to be taken, some decson has to be mposed aganst someone s wll. Guneer 1994 argues that super-majortaran votng may be one means to protect mnorty rghts. Ths s somewhat ronc gven that super-majortaran rules, such as the flbuster n the Senate, have been hstorcally employed to obstruct cvl rghts legslaton. Whle Guneer s certanly correct to pont out that super-majortaran decson rules are wdespread, t not clear that the excluson of mnortes Guneer seeks to remedy results from majorty rule, as much as from certan wnner-take-all nsttutons such as sngle member dstrct electons. Guneer s supportve of proportonal representaton. Mller 1996 provdes a socal choce theoretc analyss of Guneer s clams. Our thnkng about majorty rule has been consderably sharpened by formal socal choce theory. As already noted, May 1952 shows that majorty rule s the only postvely responsve decson rule satsfyng anonymty and neutralty. In a smlar ven, Rae 1969 and Taylor 1969 show that majorty rule s the decson rule that mnmzes the probablty that an agent votes for somethng that s not enacted or votes aganst somethng that s. Straffn 1977 shows that majorty rule s the decson rule that maxmzes responsveness to ndvdual preferences. Socal choce theory has demonstrated that majorty rule s prone to cycles ondorcet 1788/1955; Arrow 1952; Plott 1967; McKelvey 1976, 1979; Schofeld The condtons under whch ths apples to super-majortaran decson rules has been explored by Nakamura 1979, Greenberg 1979 and Schofeld et al Mller 1980 and McKelvey 1986 show there are strct lmts to majorty rule cyclng under normal nsttutonal settngs, and Mller 1983 argues that the nstablty produced by cyclng may actually benefcal to systemc stablty by assurng that there are no permanent losers. ang and Slotznck 1987 show that under super-majortaran rules, t may be strategcally ratonal for blockng mnortes to defend extreme status uo postons, even though they would lke to see them replaced. Although t s tangental to our concerns here, there s some lterature on the effect of super-majortaran rules on economc outcomes. As noted, Buchanan and Tullock 1962 argue that only unanmty guarantees economc effcency. Barry 1965, however, argues that the use of the offensve veto may lead to economc neffcency groups wth veto power may try to use that veto to extort prvleges, whch may lead to worthwhle projects not beng undertaken. Moe 1992 argues that super-majortaran nsttutons n ongress lead to neffcent bureaucraces, n that bureaucratc structures are desgned not to maxmze the performance of an agency, but to lock n the gans of the wnnng coalton and prevent future ongressonal majortes and admnstratons from beng able to change the goals of the agency. Huber et al fnd that the presence of external checks and balances have a consderable effect on the sze and type of welfare state a country adopts.

6 4 Pathologes Of Super-Majorty Rule Usng some smple examples, we can llustrate some of the problems that super-majortaran rules can produce. Such rules can lead to the complete excluson of mnortes, to mmoblsm where the status uo s mpossble to challenge, to stuatons where deologcally concentrated mnortes are advantaged over more dspersed majortes, and even to stuatons where ponts at the very extremes are strategcally defended by blockng coaltons. a 3 b 3 c 2 Fgure 1. onfguraton of Eght Voters onsder the stuaton depcted n Fgure 1. There are eght voters wth smple spatal utlty functons they each prefer outcomes closer to ther deal pont. Three have deal ponts at poston a, three at poston b and two at poston c. Under majorty rule fve votes are reured to defeat a proposal, and thus there s no core a proposal or set of proposals that cannot be defeated. For any proposal t s possble to fnd a counter-proposal that fve voters prefer. However, the voters at postons a, b, c have eual barganng power n determnng the outcome. When we move from majorty rule to supermajorty rule, ths changes. If we adopt a super-majortaran uota of sx to pass a proposal, the sold lne between postons a and b becomes the core. It s mpossble to fnd a counter-proposal that sx voters prefer to a pont on the lne between a and b. As a result, the voters at c lose all barganng power and nfluence over the outcome. If we ncrease the uota from sx to seven, then any pont n the trangle abc wll be a core pont. We may characterze ths stuaton as a tyranny of the status uo. As long as the status uo s wthn abc, t cannot be changed. Whoever had nfluence when the status uo came to pass has ther way. Thus we can show that t s at least possble that adoptng super-majortaran rules may severely harm the nterests of mnortes, compared to ther poston under majorty rule. c 1 d 1 b 1 a 1 e 1 Fgure 2. onfguraton of Fve Voters

7 Of course, super-majortaran rules can advantage certan mnortes. ertanly a very hgh uota wll advantage mnortes that are favorable towards the status uo. Furthermore, super-majortaran rules may advantage deologcally concentrated mnortes. onsder Fgure 2. Here there are fve voters, at postons a, b, c, and d and e. Under majorty rule =3 there s no core. In terms of barganng power, the voters at a and b mght have an advantage, n that they can jon wth any other voter to form a majorty, but they are not able to mpose ther wll on the others. However, f we ncrease the uota to 4, then the shaded area becomes the core. The nfluence of one mnorty the two voters at poston a and b has ncreased, but at the expense of the smaller mnorty postons c, d and e. Thus we would expect super-majortaran rules to beneft mnortes who are large and concentrated enough to form blockng coaltons at the expense of smaller and less concentrated mnortes. Fgure 3 whch s based on ang and Slotznck 1987 llustrates an even more problematc stuaton that can arse under super-majortaran rules. Here status uo pont S may be mpossble to overturn wth a votng uota of four, even though t s not n the core. If we have a super-majortaran rule wth a uota of four, then the shaded area n the fgure s the core. If a pont n the core s the status uo, t wll be mpossble to overturn t. However, f we start from status uo pont S, t may be that we never get to the core. Ths s because voters a and e form a blockng coalton. There are ponts that four voters would prefer to pont S. However, there s no pont n the core that ether voter a or e prefers to the status uo, as can be seen from the fact that the so-utlty curves of a and e do not ntersect the core. As a result t may be strategcally ratonal for a and e to block any attempt to move from the status uo. Although there are ponts that they prefer to the status uo, f these are adopted, ths may lead to the adopton n the next round or later of core ponts that voters a and e do not prefer to the status uo S. In an expermental settng, ang and Slotznck 1987 confrm the exstence of ths phenomenon. The stuaton here may appear famlar. A blockng coalton defends a status uo that some of ts members mght lke to change, because t fears that f t allows change, ths wll open the floodgates to further changes that t vews as undesrable. 5

8 6 a 1 b 1 S status uo c 1 ore wth =4 e 1 d 1 Fgure 3. onfguraton of Fve Voters wth =4 and Status Quo Pont S Analyzng Super-Majorty Rule We have seen that super-majortaran rule can have some problematc effects. Ths secton wll show why super-majorty rule s nherently less democratc than majorty rule n that t uses less of the nformaton we have about the preferences of socety. When we move from majorty to super-majorty rule, we effectvely throw away the preference nformaton we have about pars of alternatves where a super-majorty does not prefer ether alternatve to the other. In place of usng preference nformaton to decde between the two alternatves, we have to decde on some other ground. Usually, the alternatve that s the status uo prevals. Thus the decson rule s based n favor of one of the alternatves, or n the language of May s theorem, s not neutral. We have replaced a democratc decson wth an authortatve one; a decson based on preferences wth a decson based on precedent. Of course, as we move from majorty to super-majorty rule, we reduce the nstablty n outcomes assocated wth majorty rule. Ths s precsely because we are decdng between fewer pars of alternatves usng socety s preferences. Instablty under majorty rule results from the fact that socal preferences may nvolves cycles a s preferred to b s preferred to c s preferred to a. As we stop relyng on preferences and rely more on whch alternatve s the

9 7 status uo, the probablty of such cycles dmnshes. However, n the next secton, I wll show that t s precsely ths nstablty that offers protecton to mnortes under majorty rule. In secton 6, I wll argue that nstablty n outcomes actually enhances systemc stablty, followng the argument n Mller We wll analyze majorty rule and super-majortaran decson rules usng the graphtheoretc framework from Mller et us assume that there are an odd number of voters and that the preferences of all voters over possble alternatves are a lnear orderng. That s, f 1 and 2 are alternatves, then voter a ether prefers 1 to 2, or 2 to 1; furthermore voter a s preferences are transtve f voter a prefers 1 to 2, and 2 to 3, then voter a prefers 1 to 3. Then Mller 1980 shows that socety s preferences under majorty rule can be expressed as a tournament a complete asymmetrc drected graph. If we take any two alternatves 1 and 2, then ether socety prefers 1 to 2 or 2 to 1. However, due to reasons famlar from Arrow 1952 we cannot expect socety s preferences, n general, to be transtve. Ths s llustrated n Fgure 4 a. A lne from a pont 4 to pont 1 means that alternatve 4 s majorty rule preferred to alternatve 1. All ponts are connected, ndcatng that socety has a strct preference over all pars of alternatves. However, socety s preferences are not transtve alternatves 3, 4, and 5 form a cycle a b Fgure 4. Socety s Preferences over Alternatves under Majorty and Super-majorty Rule et us consder what happens when we go from majorty to super-majortaran rule, by rasng the votng uota reured for one alternatve to beat another from 50%+1 to some hgher value. As the votng uota s rased, some of socety s preference relatons are erased. If the votng uota s, and the proporton of the populaton that prefers alternatve 3 to 4 s greater than 50%, but less than, then the decson rule no longer gves a socal preference between alternatves 3 and 4. Thus we go from the stuaton depcted n Fgure 4 a, where the decson rule ranks all pars of alternatves, to that n Fgure 4b, where the dotted lne represent preference relatons that have been erased. As we further ncrease the votng uota, preference relatons cannot be replaced. If people do not prefer alternatve 1 to 2, or 2 to 1, then a hgher uota * clearly cannot prefer 1 to 2 or vce versa. Thus as we ncrease the votng uota we make the decson rule less responsve to the preferences of the populaton. When the uota = 50% + 1, then the decson rule regsters any majorty preference as a socal preference. When we rase the uota to *, then majorty preferences that are less than * are dsregarded. As such, the hgher we rase the uota, the less of the nformaton about majorty preferences we use. However, as Rae 1975 argues, t may well be necessary to make a decson between two alternatves, even though our decson rule does not rank them. In ths case some other

10 8 crteron, outsde of the populaton s preferences, has to be used. Usually n the case of super-majortaran systems, the status uo s prvleged. The alternatve that s the status uo s mantaned unless some other alternatve s preferred by uota of the populaton. Ths, however, has sgnfcant dstrbutonal conseuences. Under the super-majortaran uota, every blockng coalton sze 1- can block any alternatve that makes t worse off than the status uo. The status uo becomes essentally the guaranteed poston of each blockng coalton. As a result a great deal of the dstrbutonal barganng that otherwse would go on, may be pre-empted. The decson s based n favor of the status uo, and thus by extenson n favor of those partes who favor or can tolerate the status uo. Smlarly, we can see why ncreasng the votng uota reduces the nstablty assocated wth majorty rule. Saar 1997 shows that as we ncrease the votng uota, then the set of alternatves that are not defeated by any other alternatve the core must expand monotoncally. That s, the core under votng uota * must be a subset of the core under uota **, f ** *. onsderng Fgure 4, we can see why ths must be the case. As we rase the uota from * to **, we erase the lnes representng the preference relatons between any two alternatves where the majorty n favor of one over the other s less than **, but greater than *. However, no new socal preference relatons are added f the socal preference between two alternatves s undefned under uota *, t wll stll be undefned under the hgher uota **. As a result, the set of alternatves that are undefeated may ncrease as some socal preference relatons are deleted. However, any alternatve that s undefeated under uota * wll also be undefeated under uota **, as no new socal preference have been added. Nakamura 1979 gves condtons for a smple game such as our votng rule to not have cycles. Applyng ths result to a super-majortaran votng system, we fnd that the votng systems wll produce no cycles under any preference profle when the uota n p- 1 / p, where n s the populaton sze and p s the number of alternatves. If preferences and alternatves can be represented spatally, then a core must exst f n w-1 / w, where w s the number of dmensons Greenberg 1979; Schofeld et al In the case of unanmty, transtvty s ensured wth any number of alternatves or dmensons. If alternatve 1 s unanmously preferred over alternatve 2, and alternatve 2 s unanmously preferred over alternatve 3, then alternatve 1 must be unanmously preferred over 3, because we assume that every ndvdual s preferences are a lnear orderng. However, we acheve ths stablty by elmnatng a great deal of nformaton about socety s preferences. onsder Fgure 5, whch represents hypothetcal socal preference relatons under unanmty. ompared to Fgure 4, there are few lnes connectng the nodes. The only lnes remanng are those between one alternatve that s unanmously preferred to another. Thus alternatve 1 s undefeated that s to say, t s n the core and s Pareto-optmal there s no alternatve that s unanmously preferred to t, even though t s defeated under majorty rule by alternatves 3 and 5, whch are not n the core Fgure 5. Socal Preference Relatons under Unanmty

11 9 Thus we can see that the effect of gong from majorty rule to more supermajortaran uota rules s to gnore more and more preference nformaton and to rely more and more on precedent whchever alternatve s establshed as the status uo prevals. Naturally, ths makes the status uo more stable. However, as I wll show n the next secton, t s precsely ths nstablty the ablty to overturn undesrable outcomes f necessary that guarantees protecton to mnortes. Super-Majorty Rule And The Protecton Of Mnortes It has freuently been argued that super-majortaran decson rules both explct and mplct safeguard mnortes. However, we have seen that super-majortaran processes essentally dscard preference nformaton when the majorty s less than uota, and mposes the status uo n these cases. To argue that replacng margnal democratc outcomes wth a pror outcomes protects mnortes reures some strong assumptons. Frst, we must beleve that mnortes are more at threat from a change n the status uo than from a falure to change the status uo, ether n response to an exstng njustce or some new threat. Second, we must be able to say what the status uo s and wll be. In other words, we reure certanty or at least a hgh degree of confdence about what the status uo wll actually be n the future. The next secton argues that ths assumpton of certanty s unrealstc over the tmespan of consttutonal arrangements. Ths secton shows that f we do not have pror knowledge of our nterests, majorty rule provdes most protecton to the worst-off mnorty. Suppose we are choosng a votng system behnd a vel of gnorance of the type proposed n Rawls 1971/1999. That s to say, we have no partcular nformaton, and thus are not aware of our preferences or the status uo pont. Under these condtons Rawls argues that t s ratonal to employ the maxmn prncple we choose the rule that maxmzes our utlty n the worst stuaton we could fnd ourselves. The worst outcome we could fnd ourselves n would be to be faced wth a very unfavorable outcome that we are unable to overturn by jonng wth a coalton of other voters. onsderng dfferent votng uotas, we can show that the hgher the uota, the lower the utlty floor we are guaranteed, and thus that the system that guarantees us the hghest utlty floor s majorty rule: Proposton 1: Gven a votng rule wth uota, let m be the utlty assocated wth the least preferred poston for agent that, f enacted, no coalton ncludng could overturn gven sncere votng. Then m m +1. Proof n appendx The ntuton behnd the proof s straghtforward. et us defne the ore the core for voter as the set of alternatves that voter cannot overturn by jonng wth a coalton other voters of sze or greater and replacng t wth another alternatve. The worst thng that can happen to voter s the outcome n the ore that s least favorable to voter. As the uota rule ncreases, the sze of the ore monotoncally ncreases, as some socal preference relatons are deleted and none are added, so some alternatves that voter prevously could have overturned are now nvulnerable. As the ore under uota * must be a subset of the ore under uota ** ** > *, so the worst outcome for voter n the ore under uota

12 10 ** must be at least as bad as the worst outcome under uota *, and possbly worse. Thus a hgher uota exposes voter to potentally worse outcomes that cannot be overturned through a coalton wth other voters. Majorty rule s the decson wth the lowest votng uota that does not result n ndetermnate outcomes. A uota of less that 50% can result n stuatons where alternatve 1 s socally preferred to alternatve 2 and alternatve 2 s socally preferred to alternatve 1. Thus majorty rule offers maxmum protecton aganst the mposton of an outcome that a voter s unable to bargan to overturn. Smple majorty rule gves us the most protecton aganst havng unfavorable outcomes mposed on us. Indeed, provded that preferences are dstrbuted n at least two dmensons and do not meet some very strngent symmetry condtons Plott 1967, McKelvey 1976, 1979, Schofeld 1978 we can fnd a coalton to overturn any outcome except our deal pont. The problem s that any other agent can do the same. Whether ths forces reasonable outcomes, or whether ths leads to chaos s a behavoral ueston, whch wll be addressed below. Nevertheless, majorty rule offers greater protecton aganst an mposed outcome than any other system, and therefore the system that offers most protecton for mnorty rghts, s, roncally, majorty rule. Super-Majorty Rule and Uncertanty To argue that t s prudent to prvlege the status uo over an alternatve that s majorty preferred, t s necessary to be able to say what the status uo s. Furthermore, the greater the degree of uncertanty about the status uo, the more reasonable t s to use the Rawlsan assumpton of a vel of gnorance as a devce for argung about justce. I wll argue here that over the tmeframes relevant to consttutonal choce, the status uo may be ute ndetermnate. The analyss her bulds on the work of Brunel-Petron concernng the theory of rghts. Brunel-Petron argues that we have to consder rghts as clams on outcomes. The mappng, however, between the law and the rghts we possess n practce s problematc. Ths mappng may change over tme, n partcular n responses to changes n technology and socal mores. Thus although the law may not change, f there s a sgnfcant change n technology or mores, ths law may represent a very dfferent outcome. Ths dea can be appled to our consderaton of the status uo. Indeed between the set of possble status uo postons and the set of possble pay-offs, there are several mappngs. Super-majortaran rule prvleges the status uo by makng t hard to change the law. The status uo that s protected, however, s not an outcome or pay-off, but rather a set of legal formalsms. If the way n whch these legal formalsms are translated nto outcomes or pay-offs changes, then the substantve status uo wll change, although the formal status uo s untouched. Fgure 6 llustrates the mappngs from the set of possble laws to the set of possble pay-offs. Frst, the wrtten law produced by legslaton has to be translated nto government acton or polcy. The nterpretaton of law by the executve and the courts, can, of course, change over tme, whch wll substantvely change polcy. Ths knd of slppage s partcularly sgnfcant n super-majortaran systems. Under smple majorty rule, t s relatvely easy for the legslature to correct changes n nterpretaton by the executve or judcary by smply passng new legslaton. If the decson rule s super-majortaran, however, the executve and judcary may have far more dscreton. As such, super-

13 11 majortaran rule may amount n practce to a form of concealed guardanshp, f the nterpretaton s performed by an unelected body See Dahl 1956 for a crtue of the US Supreme ourt n ths role. Moe and Howell 1999 argue that super-majortaran rules effectvely empower the US Presdent because he has some unlateral power to determne how laws are mplemented. kewse t can be argued that the power of the European ommsson the executve of the European Unon s enhanced by the fact that the ouncl of Mnsters proceeds on the bass of unanmty or ualfed majorty votng, and thus has a hard tme overturnng ommsson rulngs. Technology and socal mores Interpretaton Polcy Outcomes Preferences The aw legal formalsms Pay-offs Fgure 6. Mappngs from the aw to Fnal Pay-offs Even f government polcy remans the constant over tme, the outcomes that ths polcy represents may change because of changes n technology, socal mores or the envronment n general. For example, slow government response may be tolerable or even desrable n normal condtons, but dsastrous n tme of natonal emergency. Ths s why Federalst 22 s so hostle to super-majortaransm. Technology may also affect the effectve conseuences of laws. The rght to keep weapons may have very dfferent conseuences dependng on the development of mltary technology. The development of effcent computers has surely changed the mpact of data prvacy laws or lack thereof. Smlarly, socal mores change the conseuences of laws. A law bannng publc nudty would have no effect f nobody wshed to behave n ths way, or f nobody mnded. Nether would the lack of such a law. In addton to the drect effect on the relatonshp between polcy and outcomes that a change n technology or mores may have, there may be ndrect effects medated through economc processes. hanges n technology change producton functons. As a result, demand for factors of producton change, as does ther relatve value. The factors that are crucal to people s lvelhood and welfare, and the relatonshp between them, also change. A type of economc regulaton that appeared eutable may no longer be so. In an agraran socety, we would not expect people to even conceptualze the need for organzed labor. It s only wth the advent of an ndustral socety that dsputes over the rght to organze or not to be organzed become salent, as ndustral labor s now a key factor of producton. ack of specfc laws dealng wth organzed labor have no effect when there s no socetal drve to organzed labor, but may later become crucally mportant. Smlarly, n a post-ndustral

14 12 economy, ntellectual captal may become the crucal factor of producton. aws concernng ntellectual property that seemed reasonable when employees were expected to works for a frm for lfe, may be extremely problematc n a socety where freuent job shfts are commonplace. Suddenly, the dstrbuton of ntellectual property rghts between employer and employee becomes a crucal concern. Fnally, even f outcomes reman constant over tme, socety s preferences over these outcomes may change. Thus the eventual pay-offs to the varous actors would be dfferent, and our consderatons of justce would have to adapt to ths. One partcular nstance n whch socety s preferences may be exceptonal s durng tme of crss. It s possble that legslaton may be passed hastly under crss condtons despte super-majortaran rules, because all partes may want acton of some type to be taken uckly. However, these hastly made decsons may prove extremely dffcult to change when tme allows more detaled consderaton. It s very dffcult to argue that we can know what the status uo wll be n the medum term, let alone that t wll be a satsfactory outcome. Thus attempts to engneer specfc outcomes usng consttutonal mechansms appear hubrstc. Furthermore, we have not yet consdered one further change we would expect to occur over tme: The agents themselves wll change, as some de and others are born. Super-majortaran rule prvleges the status uo, and as such the choces of one generaton over that the succeedng one. In ths sense super-majortaran rule can be thought of a mechansm by whch a domnant group today protects tself aganst the majorty of the next generaton, beueathng to ther chldren a world that not only dd they not create, but that they may not even be able to revse. Stablty and The Benefts of yclng The nstablty resultng from cyclng s the other man objecton to majorty rule gven as justfcaton for the adopton of super-majortaran decson rules. However, followng Mller 1983 I have argued that t s precsely ths nstablty that allows us to overcome the problem of the protecton of mnortes from a tyranny of the majorty. We have seen that super-majortaran rule merely replaces the possblty of domnaton by a majorty wth domnaton by a prvleged mnorty. However, the presence of multple, cyclng majortes provdes the possblty of a check on majortes wthout artfcally empowerng a mnorty. Furthermore, we now have theoretcal and emprcal reasons to beleve that there are strct bounds on the nstablty assocated wth majorty rule. Not least amongst these s that fact that countres wth majorty rule legslatures wth few f any external checks and balances such as the Netherlands, Austra and Sweden do not appear partcularly unstable. Indeed these are amongst the countres dentfed by jphart 1999 as beng consensual. Majorty rule cyclng can protect mnortes n two ways. Frst, as Mller 1983 argues, t can ensure that there are no permanent losers. A group out of power can always expect to be able to defeat the ncumbents n the future, and thus have an ncentve to keep playng the game, whch enhances the stablty of the system. Second, the need to buld and mantan majorty coaltons encourages ncluson. Any party that s excluded and feels lke the nterests of ts consttuents are threatened, can undermne the governng coalton by offerng ts support to one of the governng partes at a very low cost. By offerng some of the governng partes outcomes they fnd very desrable n exchange for only the outcome t vews as absolutely vtal, an excluded party can break up the governng coalton. Of course,

15 13 ths outcome s not stable, as the new excluded partes have an ncentve to pursue the same strategy. The only way to acheve stablty s to organze outcomes so that excluded partes are ndfferent to stayng n opposton and tryng to break nto the governng coalton. The fact that cyclng majorty rule makes t easy for an excluded party to make trouble by tryng to undermne the governng coalton, encourages power sharng. Ironcally, majorty rule seems to encourage the knd of takng turns solutons that Guneer 1994 advocates. Ths provdes a soluton to the problem of ntensty of preferences, whch Dahl 1956, ch.4 suggested was nsoluble. If a majorty and a mnorty dsagree on an ssue whch both vew as beng vtal to ther survval, there s probably no soluton satsfactory to both partes. However, n a stuaton where a mnorty vews an ssue as vtal, whle the majorty s relatvely apathetc about t, sngle-ssue majorty rule produces an unsatsfactory result. However, any attempt to safeguard the mnorty wth specal protectons s problematc, as there s no way to measure what ssues actually are vtal to the mnorty, and t wll have an ncentve to exaggerate n order to get ts way on more ssues. However, a majorty rule parlament n whch multple ssues are consdered provdes a soluton. The mnorty can trade votes on ssues t consders of mnor mportance for support on ssues t consders vtal. Interestngly, although Buchanan and Tullock 1962 have been amongst the foremost advocates of unanmty rule, the followng uotaton alculus of onsent, p. 132 provdes an argument very smlar to the one above: Applyng the strct Pareto rules for determnng whether one socal stuaton represents an mprovement over another, almost any system that allows some such exchange to take place would be superor to that system whch weghts all preferences eually on each ssue. By way of llustraton, t s concevable that a proposal to prohbt Southern Democrats from havng access to free rado tme mght be passed by smple majorty vote n a natonal referendum should the ssue be rased n ths way. Such a measure, however, would not have the slghtest chance of beng adopted by the decson-makng process actually prevalng n the Unted States. The measure would never pass the ongress because the supporters of the mnorty threatened wth the damage would, f the ssue arose, be wllng to promse support on other measures n return for votes aganst such dscrmnatory legslaton. Whle cyclng has benefcal effects n terms of the protecton of mnortes, the negatve effects n terms of nstablty have been overstated. On the theoretcal level, t s now apparent that the presence of cycles does not mean chaos n the sense of anythng can happen, and probably wll. Early nterpretatons of McKelvey s 1976, 1979 and Schofeld s 1978 global cyclng results notably Rker 1982 argued that the results had a corrosve effect on democratc theory, makng any outcome as vald as any other, and makng agenda-settng poltcans the key determnant of outcomes. However, t s now clear that even wthout a core, majorty rule s rather more orderly. For example, Mller 1980 shows that n a varety of nsttutonal settngs, outcomes wll tend to fall n the uncovered set the set of ponts that are not covered, n that there s no other pont that beats them and everythng that they beat. McKelvey 1986 shows that the uncovered set s typcally a relatvely small, central set. Furthermore, t can be shown that any pont outsde the uncovered set can be beaten by any pont n the generalzed medan set. In order to produce

16 14 an outcome beyond the uncovered set would reure a Mephstolean agenda manpulator who s able to permanently exclude alternatves around the medan from consderaton. 8 Expermental research also suggests that even where there s no core, majorty rule tends to produce outcomes n a small central set, smlar to the uncovered set see McKelvey and Ordeshook There s no evdence that countres that have majorty rule parlaments and very few checks and balances are partcularly unstable. 9 Indeed, many of the small European democraces dentfed by jphart as beng consensual are of ths type, wth majorty rule legslatures elected by proportonal representaton Netherlands, Belgum, Austra, Sweden, Norway, Denmark. 10 Gven the presence of cross-cuttng socal cleavages, and the fact that there are certanly multple ssue dmensons, we would expect cyclng majortes amongst partes. Indeed, n all of these countres we have seen government by dfferent combnatons of partes, and by combnatons that do not make sense n terms of a one-dmensonal leftrght deologcal dvson. It s true that these countres have varous nsttutons that facltate power sharng. However, these nsttutons exst wthn the context of majorty rule. There have been grand coaltons and delegaton of decson makng to communtes and functonal bodes. Even when some partes are excluded from the governng coalton, they are not completely excluded from decson makng, as parlamentary commttees and varous consultatve councls are mportant. As Strom 1992 argues the power dfference between those n government and those out s small. There s a wllngness by party leaders to take decsons n an nclusve manner, as well as a keen awareness that those excluded are part of an alternatve coalton. Fnally we may speculate that majorty rule s rather conducve to consensual outcomes. Although jphart 1977, 1984, 1999 does not talk n the language of socal choce theory, hs model of consensus democracy seems to be drven by the presence of cyclng majortes. onsensual outcomes are more lkely when no group s able to domnate on ts own, but nstead groups have to reach accommodatons wth each other. In these crcumstance majorty rule has the advantage of makng ntransgence a very rsky strategy. If I refuse to compromse, you can reach a deal wth someone else and I wll be excluded. Wth super-majorty rules, however, as I have argued above, ntransgence by a mnorty may often be an optmal strategy, as a mnorty may have the power to block changes and thus have somethng to defend. As such consensus unanmty may be the decson rule least lkely to produce consensual behavor. Rather than power sharng beng a response to protect mnortes from majorty rule, t may actually be the type of response that majorty rule encourages. oncluson Dahl 1956 argues that majorty rule unuely embodes the values of popular soveregnty and poltcal eualty. However, t s argued by Dahl, amongst others, that these are not the only values we are concerned wth n matters of consttutonal choce. The two alternatve values most commonly cted are the protecton of mnorty rghts and stablty. It s commonly argued that there s a trade-off between poltcal eualty best served by majorty rule on one hand, and mnorty protecton and stablty best served by external checks and balances on the other. Ths paper has argued that ths trade-off s llusory, and that majorty

17 rule provdes more protecton for the worst-off mnorty than any other decson rule. Furthermore, t does so precsely because of the nstablty nherent n majorty rule. Majorty rule offers most protecton to mnortes because t makes t easest for a mnorty to form a coalton that can overturn an unacceptable outcome. Super-majorty rules can certanly protect or rather prvlege some mnortes, but only at the expense of others. It s not logcally possble for every mnorty to be prvleged over every other mnorty. Super-majorty rules make the status uo hard to overturn and thus prvlege mnortes who favor the status uo over those who favor changng t. Arguments n favor of supermajortaran nsttutons have tended to be bult on the assumpton that the threat to rghts from government acton or a change n the law s greater than the threat from government nacton or the mantenance of current laws. Gven the hstory of the Unted States ths assumpton s problematc, especally gven the use of super-majortaran nsttutons to mpede the extenson of cvl rghts. Furthermore, gven uncertanty about legal nterpretaton, technology, socal mores and preferences over the tmescale nvolved n consttutonal choce, any assumptons about where the threat to rghts are lkely to le are nevtably heroc. Whle super-majortaran rules are only able to protect some mnortes at the expense of others, the nstablty resultng from global cyclng under majorty rule offers an alternatve approach to the problem of the tyranny of the majorty. The costs of nstablty resultng from cyclng have been overstated theoretcally we no longer expect unrestrcted or chaotc outcomes, and the countres that practce relatvely unchecked majorty rule are ute stable. Nevertheless, the possblty of cyclng seems to lead to nclusve poltcs, n that t s always necessary to assemble a broad coalton, and any coalton can be splt. There s no tyranny of the majorty because there s no sngle, cohesve majorty ready to domnate everyone else. Ths, of course, s essentally the extended republc argument made by James Madson at the onsttutonal onventon of 1787 and n Federalst

18 16 Bblography Aldrch, J Why Partes? The Orgn and Transformaton of Poltcal Partes n Amerca. hcago: Unversty of hcago Press. Barry, B. 1965/1990. Poltcal Argument. New York: Harvester Wheatsheaf. Brunel-Petron ontrbuton à l analyse des drots en theore du chox socal. Thèse pour le doctorat de 'Unversté de aen, specalté scences economues. Buchanan, J. and Tullock, G The alculus of onsent: ogcal Foundatons of onsttutonal Democracy. Ann Arbor: Unversty of Mchgan Press. alhoun, J. 1842/1982. On the Veto Power. In Krk, R. ed. The Portable onservatve Reader. New York:Pengun Books. alhoun, J. 1850/1943. A Dsuston on Government. New York: P. Smth. ondorcet, Marus de de artat, M /1995. On the onsttuton and Functon of Provncal Assembles. In Mclean, I. and Urken, A. eds. lassc of Socal hoce. Ann Arbor: Unversty of Mchgan Press. Dahl, R A Preface to Democratc Theory. hcago: Unversty of hcago Press. Dahl. R Democracy and ts rtcs. New Haven: Yale Unversty Press. Greenberg, J onsstent Majorty Rules over ompact Sets of Alternatves. Econometrca 47.3: Guner, The Tyranny of the Majorty: Fundamental Farness n Representatve Democracy. NY: The Free Press. Hamlton, A., Madson, J. and Jay, J. 1788/1961. The Federalst Papers. New York: Mentor. Huber, E., Ragn,. and Stephens, J. Socal Democracy, hrstan Democracy, onsttutonal Structure and the Welfare State. Amercan Journal of Socology 99.3: ang, J and Slotznck, B Vable Alternatve to the Status Quo. Journal of onflct Resoluton 31: jphart, A Democracy n Plural Socetes. Yale Unversty Press. jphart, A Democraces: Patterns of Majortaran and onsensus Government n Twenty-One ountres. New Haven: Yale Unversty Press. jphart, A Patterns of Democracy: Government Forms and Performance n Thrty-Sx ountres. New Haven: Yale Unversty Press. Madson, J. 1840/1966. Notes of Debates n the Federal onventon of 1787 Reported by James Madson. Athens, OH: Oho Unversty Press. Madson, J Wrtngs. New York: The brary of Amerca. May, K A Set of Independent Necessary and Suffcent ondtons for Smple Majorty Decson. Econometrca 20: McKelvey, R Intranstvtes n Multdmensonal Votng Models and Some Implcatons for Agenda ontrol. Journal of Economc Theory 16:

19 17 McKelvey, R General ondtons for Global Intranstvtes n Formal Votng Models. Econometrca: McKelvey, R overng, Domnance, and the Insttuton-Free Propertes of Socal hoce. Amercan Journal of Poltcal Scence 30.2: McKelvey, R A Decade of Expermental Research on Spatal Models. In Enelow, J and Hnch, M., Advances n the Spatal Theory of Votng. ambrdge Unversty Press. Mller, N A New Soluton Set for Tournaments and Majorty Votng: Further Graph- Theoretcal Approaches to the Theory of Votng. Amercan Journal of Poltcal Scence 24.1: Mller, N Socal hoce and Pluralsm. Amercan Poltcal Scence Revew 77.3: Mller, N Majorty Rule and Mnorty Interests. In Shapro, I. and Hardn, R. eds. Poltcal Order: Nomos XXXVIII. New York: New York Unversty Press. Moe, T The Poltcs of Bureaucratc Structure. In hubb, J. and Peterson, P. an the Government Govern? Washngton, D: The Brookngs Insttuton. Moe, T. and Howell, W The Presdental Power of Unlateral Acton. The Journal of aw, Economcs and Organzaton 15: Nakamura, K The Vetoers n a Smple Game wth Ordnal Preferences. Internatonal Journal of Game Theory 8: Plott, A Noton of Eulbrum under Majorty Rule. Amercan Economc Revew 57: Rae, D Decson-Rules and Indvdual Values n onsttutonal hoce. Amercan Poltcal Scence Revew 63.1: Rae, D The mts of onsensual Decson Makng. Amercan Poltcal Scence Revew 69.4: Rawls, J. 1971/1999. A Theory of Justce. Harvard Unversty Press. Rker, W beralsm versus Populsm: A onfrontaton Between the Theory of Democracy and the Theory of Socal hoce. San Francsco: Freeman Saar, D The Generc Exstence of a ore for Q-Rules. Economc Theory 9: Schofeld, N Instablty of Smple Dynamc Games. Revew of Economc Studes 45.3: Schofeld, N., Grofman, B. and Feld, S The ore and Stablty of Group hoce n Spatal Votng Games. Amercan Poltcal Scence Revew 82.1: Sen, A ollectve hoce and Socal Welfare. San Francsco: Holden-Day. Straffn, P. Jr Majorty Rule and General Decson Rules. Theory and Decson 8: Strom, K Mnorty Government and Majorty Rule. ambrdge: ambrdge Unversty Press. Tataru, M Growth Rates n Multdmensonal Spatal Votng. Mathematcal Socal Scence 37: Taylor, M rtue and omment: Proof of a Theorem on Majorty Rule. Behavoral Scence 14: van der Hout, E Axoms haracterzng the Pluralty Rankng Rule. Paper presented at the annual meetng of the Publc hoce Socety, San Dego, A, March