George Tsebelis Veto Players

8/12/2019 George Tsebelis Veto Players

1/38

Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentarism,Multicameralism and MultipartyismAuthor(s): George TsebelisSource: British Journal of Political Science, Vol. 25, No. 3 (Jul., 1995), pp. 289-325Published by: Cambridge University PressStable URL: http://www.jstor.org/stable/194257

Accessed: 29/09/2010 14:53

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=cup.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

Cambridge University Pressis collaborating with JSTOR to digitize, preserve and extend access toBritish

Journal of Political Science.

http://www.jstor.org
http://www.jstor.org/action/showPublisher?publisherCode=cuphttp://www.jstor.org/stable/194257?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=cuphttp://www.jstor.org/action/showPublisher?publisherCode=cuphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/194257?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=cup


2/38


3/38


4/38

Decision Making in Political Systemsecision Making in Political Systemspresidentialism suffers from such negative factors as temporal rigidities,majoritarian tendencies and dual democratic legitimacies.'? Arguments overbicameralism closely parallel the arguments on presidentialism by focusing onchecks and balances versus dual democratic legitimacies. Finally, two-partysystems are thought to provide moderation of parties, stable executives, clearchoices and responsible majorities. But Lijphart systematically rebuts each oneof these points; ' and Huber and Powell actually find smaller distances betweenthe median voter and the government median in multi-party systems than intwo-party systems.'2On the empirical side, analysts often compare countries which differ alonga cluster of characteristics. For example, Anglo-Saxon authors frequentlycompare the United Kingdom with the United States. But the differencesbetween these two countries are numerous: presidential vs. parliamentarysystems, bicameralism vs. (de facto) unicameralism, undisciplined vs. disci-plined parties, appointed vs. independent bureaucracies, and the presence vs. theabsence of a strong supreme court.'3 Without a theoretical model, it is difficultto sort out which of these differences are causally prior to others. Alternatively,with a small number of like cases, any particular outcome will be overdeter-mined by the relevant variables. For example, Linz attributes the breakdown ofdemocracy in Chile to the country's presidential system,'4 whereas Horowitzargues it was due to the plurality electoral system in use in presidential electionsthere. 5 This problem can be corrected by increasing the sample size to includea large number of countries,16 or preferably the universe of relevant countries.17From this short and incomplete account of extensive literatures, I want to

10Shugart and Carey, Presidents and Assemblies, p. 29.l Lijphart,Democracies.12JohnD. Huber andG. Bingham Powell, 'Congruencebetween Citizens andPolicymakersin twoVisions of Liberal Democracy', World Politics, forthcoming.'3 The most famous authors that used this approachareWalterBagehot, TheEnglish Constitution

(London:ChapmanandHall, 1867); and Woodrow Wilson, Congressional Government(Gloucester,Mass.: Peter Smith, 1973 (first edn, 1885)). For a recent article using the United Kingdom and theUnited States as representativesof parliamentaryand presidential systems, see Terry M. Moe andMichael Caldwell, 'The Institutional Foundations of Democratic Government: A Comparison ofPresidential and Parliamentary Systems', Journal of Institutional and Theoretical Economics,150 (1994), 171-95.14Linz, 'The Perils of Presidentialism'.15Horowitz, 'Comparing Democratic Systems'.16See Lijphart,Democracies; G. Bingham Powell, ContemporaryDemocracies: Participation,Stabilityand Violence (Cambridge,Mass.: HarvardUniversity Press, 1982); KaareStrom, MinorityGovernmentand Majoritx Rule (New York: Cambridge University Press, 1990); and Shugart andCarey, Presidents and Assemblies.17 For examples of bias introduced by case selection on the dependent variable, see BarbaraGeddes, 'How the Cases You Choose Affect the Answers You Get: Selection Bias in ComparativePolitics', Political Analysis, 2 (1990), 131-49. However, even the increase of sample size does notcorrect for a bias due to the selection of existing cases from a population of possible cases withdifferent characteristics (see Adam Przeworski and FernandoLimongi, 'Selection, Counterfactualand Comparisons' (mimeo, University of Chicago); and for an empirical example along these lines

presidentialism suffers from such negative factors as temporal rigidities,majoritarian tendencies and dual democratic legitimacies.'? Arguments overbicameralism closely parallel the arguments on presidentialism by focusing onchecks and balances versus dual democratic legitimacies. Finally, two-partysystems are thought to provide moderation of parties, stable executives, clearchoices and responsible majorities. But Lijphart systematically rebuts each oneof these points; ' and Huber and Powell actually find smaller distances betweenthe median voter and the government median in multi-party systems than intwo-party systems.'2On the empirical side, analysts often compare countries which differ alonga cluster of characteristics. For example, Anglo-Saxon authors frequentlycompare the United Kingdom with the United States. But the differencesbetween these two countries are numerous: presidential vs. parliamentarysystems, bicameralism vs. (de facto) unicameralism, undisciplined vs. disci-plined parties, appointed vs. independent bureaucracies, and the presence vs. theabsence of a strong supreme court.'3 Without a theoretical model, it is difficultto sort out which of these differences are causally prior to others. Alternatively,with a small number of like cases, any particular outcome will be overdeter-mined by the relevant variables. For example, Linz attributes the breakdown ofdemocracy in Chile to the country's presidential system,'4 whereas Horowitzargues it was due to the plurality electoral system in use in presidential electionsthere. 5 This problem can be corrected by increasing the sample size to includea large number of countries,16 or preferably the universe of relevant countries.17From this short and incomplete account of extensive literatures, I want to

10Shugart and Carey, Presidents and Assemblies, p. 29.l Lijphart,Democracies.12JohnD. Huber andG. Bingham Powell, 'Congruencebetween Citizens andPolicymakersin twoVisions of Liberal Democracy', World Politics, forthcoming.'3 The most famous authors that used this approachareWalterBagehot, TheEnglish Constitution

(London:ChapmanandHall, 1867); and Woodrow Wilson, Congressional Government(Gloucester,Mass.: Peter Smith, 1973 (first edn, 1885)). For a recent article using the United Kingdom and theUnited States as representativesof parliamentaryand presidential systems, see Terry M. Moe andMichael Caldwell, 'The Institutional Foundations of Democratic Government: A Comparison ofPresidential and Parliamentary Systems', Journal of Institutional and Theoretical Economics,150 (1994), 171-95.14Linz, 'The Perils of Presidentialism'.15Horowitz, 'Comparing Democratic Systems'.16See Lijphart,Democracies; G. Bingham Powell, ContemporaryDemocracies: Participation,Stabilityand Violence (Cambridge,Mass.: HarvardUniversity Press, 1982); KaareStrom, MinorityGovernmentand Majoritx Rule (New York: Cambridge University Press, 1990); and Shugart andCarey, Presidents and Assemblies.17 For examples of bias introduced by case selection on the dependent variable, see BarbaraGeddes, 'How the Cases You Choose Affect the Answers You Get: Selection Bias in ComparativePolitics', Political Analysis, 2 (1990), 131-49. However, even the increase of sample size does notcorrect for a bias due to the selection of existing cases from a population of possible cases withdifferent characteristics (see Adam Przeworski and FernandoLimongi, 'Selection, Counterfactualand Comparisons' (mimeo, University of Chicago); and for an empirical example along these lines

29191


5/38

292 TSEBELIS92 TSEBELIShighlight one point. Recurrentthemes in all the theoretical debates include theresponsibility of elected representatives, the identifiability of decision makers,and single or dual legitimacies. However, these themes are used exclusively toexamine different variables as dichotomous pairs (regime types, legislaturetypes, partysystems); they are not used to assess the effects of combinations andhybrids, such as comparing a unicameral presidential multi-party system witha bicameral parliamentarytwo-party system.This article does not replicate the pairwise structureof these ongoing debatesseparating regime type (parliamentarismvs. presidentialism), legislature type(unicameralvs. bicameral) andpartysystem (two-partyvs. multi-party).Infact,I show that it may be misleading to examine these factors in isolation. I willargue that the logic of decision making in presidential systems is quite similarto the logic of decision making in multi-party parliamentarysystems. Similarly,bicameralism and presidentialism share common characteristics of decisionmaking. In addition, I do not aim to discuss the pros and cons of each of theinstitutional alternatives found in the title. Instead, I compare all of theseinstitutions with respect to one important variable: the capacity for policychange. My goal is to provide a consistent framework for comparisons acrossregimes, legislatures and party systems.One importantcontributionof such anapproachis thatby permittinga simpleand conceptually consistent method of making comparisons across systems, ithelps to resolve a pervasive problem of comparativepolitics: small sample size.If comparisons are permitted only across countries with the same regimetype - for example, presidential systems - then the sample size is essentiallyreduced to the Latin American countries. However, these countries also sharea host of other characteristics (economic development, party systems, partydiscipline, administrative structures, etc.), creating a serious problem ofmulticollinearity. One way of resolving this problem is to expand the samplesize by including countries thatdiffer along some of these variables. However,such an expansion requiresa theory of comparison across regime types (as wellas across party systems and legislature types), which is the purpose of thisarticle.Another purpose, and perhaps the majorcontribution of this approach, is tohelp generate hypotheses in several other areas, such as the importance andindependence of judiciaries, the independence of bureaucracies, governmentstability (in parliamentary systems) and regime stability (in presidentialsystems). Preliminaryevidence in favour of the expectations of this model willbe presented in the third part of this article.The dependent variable of my study is the potential for policy change indifferent institutional settings. I will call the absence of such potential policy(F'note continued)see George Tsebelis, 'The Power of the EuropeanParliament as a Conditional Agenda Setter', inD. Ruloff andG. Schneider, eds, Towards a New Europe: Stops and Starts in European Integration(New York: Praeger, forthcoming).

highlight one point. Recurrentthemes in all the theoretical debates include theresponsibility of elected representatives, the identifiability of decision makers,and single or dual legitimacies. However, these themes are used exclusively toexamine different variables as dichotomous pairs (regime types, legislaturetypes, partysystems); they are not used to assess the effects of combinations andhybrids, such as comparing a unicameral presidential multi-party system witha bicameral parliamentarytwo-party system.This article does not replicate the pairwise structureof these ongoing debatesseparating regime type (parliamentarismvs. presidentialism), legislature type(unicameralvs. bicameral) andpartysystem (two-partyvs. multi-party).Infact,I show that it may be misleading to examine these factors in isolation. I willargue that the logic of decision making in presidential systems is quite similarto the logic of decision making in multi-party parliamentarysystems. Similarly,bicameralism and presidentialism share common characteristics of decisionmaking. In addition, I do not aim to discuss the pros and cons of each of theinstitutional alternatives found in the title. Instead, I compare all of theseinstitutions with respect to one important variable: the capacity for policychange. My goal is to provide a consistent framework for comparisons acrossregimes, legislatures and party systems.One importantcontributionof such anapproachis thatby permittinga simpleand conceptually consistent method of making comparisons across systems, ithelps to resolve a pervasive problem of comparativepolitics: small sample size.If comparisons are permitted only across countries with the same regimetype - for example, presidential systems - then the sample size is essentiallyreduced to the Latin American countries. However, these countries also sharea host of other characteristics (economic development, party systems, partydiscipline, administrative structures, etc.), creating a serious problem ofmulticollinearity. One way of resolving this problem is to expand the samplesize by including countries thatdiffer along some of these variables. However,such an expansion requiresa theory of comparison across regime types (as wellas across party systems and legislature types), which is the purpose of thisarticle.Another purpose, and perhaps the majorcontribution of this approach, is tohelp generate hypotheses in several other areas, such as the importance andindependence of judiciaries, the independence of bureaucracies, governmentstability (in parliamentary systems) and regime stability (in presidentialsystems). Preliminaryevidence in favour of the expectations of this model willbe presented in the third part of this article.The dependent variable of my study is the potential for policy change indifferent institutional settings. I will call the absence of such potential policy(F'note continued)see George Tsebelis, 'The Power of the EuropeanParliament as a Conditional Agenda Setter', inD. Ruloff andG. Schneider, eds, Towards a New Europe: Stops and Starts in European Integration(New York: Praeger, forthcoming).


6/38

Decision Making in Political Systemsecision Making in Political Systemsstability. To paraphrase V. O. Key, a potential for policy change does notguarantee such change, but the absence of this potential precludes it.18 Policystability is different from both government stability andregime stability. Infact,as I arguein the last partof this article, they areinversely related:policy stabilitycauses government or regime instability. This analysis is based on the conceptof the veto playerin differentinstitutionalsettings. A veto playeris anindividualor collective actor whose agreement is required for a policy decision. Idemonstrate that policy stability increases with (i) the numberof veto players,(ii) their incongruence (the difference in their political positions) and (iii) theinternal cohesion of each one of them.The article is organized into three sections. Section Idiscusses the dependentvariable (policy stability) and how it can be operationalized using the conceptof 'winset', a concept taken from collective choice theory. Section IIdiscussesthe three independent variables that explain policy stability and relates them toeasily observable characteristics like regime types, the number of parties ingovernment, the number of chambers, party cohesion and other importantpolitical variables like electoral systems. Section III discusses the model'spredictions and compares it with other middle-range theories in comparativepolitics as well as some available empirical evidence.

I. POLICY STABILITY AND ITS PROXYSeveral studies correlatespecific institutionswith particular(mainly economic)outcomes. Starting with Bagehot, presidential regimes, with the diffusion ofresponsibility they entail, have been associated with high deficits.19 BothKatzenstein and Rogowski argue thatproportionalrepresentationis correlatedwith, or is conducive to, tradeopenness and economic growth.20Alternatively,Grilli et al. associate proportionalrepresentationwith high debt and inflation,and presidential systems with more responsible fiscal policies.21 Tiebout and,more recently, Weingast associate federalism with high levels of growthbecause it induces competition among constituent units.22Political scientists are often interested in the decisiveness of a politicalsystem, in other words, its capacity to solve problems when they arise. For

I8V. O. Key Jr, Politics, Parties, and Pressure Groups, 5th edn (New York: Crowell, 1964),p. 688.19Bagehot, The English Constitution.20PeterJ. Katzenstein, Small States in WorldMarkets: Industrial Policy in Europe (Ithaca,NY:Cornell University Press, 1985); and Ronald Rogowski, 'Trade and the Variety of DemocraticInstitutions', International Organization, 41 (1987), 203-23.21Vittorio Grilli, Donato Masciandaro and Guido Tabellini, 'Political and Monetary Institutionsand Public Financial Policies in the IndustrialCountries', Economic Policy, 13 (1991), 341-92.22Charles Tiebout, 'A Pure Theory of Local Expenditures', Journal of Political Economy, 64(1956), 416-24; andBarryWeingast, 'Federalismand the Political Commitmentto SustainMarkets'(mimeo, Hoover Institution, 1993).

stability. To paraphrase V. O. Key, a potential for policy change does notguarantee such change, but the absence of this potential precludes it.18 Policystability is different from both government stability andregime stability. Infact,as I arguein the last partof this article, they areinversely related:policy stabilitycauses government or regime instability. This analysis is based on the conceptof the veto playerin differentinstitutionalsettings. A veto playeris anindividualor collective actor whose agreement is required for a policy decision. Idemonstrate that policy stability increases with (i) the numberof veto players,(ii) their incongruence (the difference in their political positions) and (iii) theinternal cohesion of each one of them.The article is organized into three sections. Section Idiscusses the dependentvariable (policy stability) and how it can be operationalized using the conceptof 'winset', a concept taken from collective choice theory. Section IIdiscussesthe three independent variables that explain policy stability and relates them toeasily observable characteristics like regime types, the number of parties ingovernment, the number of chambers, party cohesion and other importantpolitical variables like electoral systems. Section III discusses the model'spredictions and compares it with other middle-range theories in comparativepolitics as well as some available empirical evidence.

I. POLICY STABILITY AND ITS PROXYSeveral studies correlatespecific institutionswith particular(mainly economic)outcomes. Starting with Bagehot, presidential regimes, with the diffusion ofresponsibility they entail, have been associated with high deficits.19 BothKatzenstein and Rogowski argue thatproportionalrepresentationis correlatedwith, or is conducive to, tradeopenness and economic growth.20Alternatively,Grilli et al. associate proportionalrepresentationwith high debt and inflation,and presidential systems with more responsible fiscal policies.21 Tiebout and,more recently, Weingast associate federalism with high levels of growthbecause it induces competition among constituent units.22Political scientists are often interested in the decisiveness of a politicalsystem, in other words, its capacity to solve problems when they arise. For

I8V. O. Key Jr, Politics, Parties, and Pressure Groups, 5th edn (New York: Crowell, 1964),p. 688.19Bagehot, The English Constitution.20PeterJ. Katzenstein, Small States in WorldMarkets: Industrial Policy in Europe (Ithaca,NY:Cornell University Press, 1985); and Ronald Rogowski, 'Trade and the Variety of DemocraticInstitutions', International Organization, 41 (1987), 203-23.21Vittorio Grilli, Donato Masciandaro and Guido Tabellini, 'Political and Monetary Institutionsand Public Financial Policies in the IndustrialCountries', Economic Policy, 13 (1991), 341-92.22Charles Tiebout, 'A Pure Theory of Local Expenditures', Journal of Political Economy, 64(1956), 416-24; andBarryWeingast, 'Federalismand the Political Commitmentto SustainMarkets'(mimeo, Hoover Institution, 1993).

29393


7/38

294 TSEBELIS94 TSEBELISexample, in a thoughtful analysis of the effects of political institutions, Weaverand Rockman distinguish:ten differentcapabilitiesthatall governmentsneed: to set and maintainprioritiesamong the many conflicting demands made upon them so that they are notoverwhelmed and bankrupted;o target resources where they are most effective;to innovate when old policies have failed; to coordinate conflicting objectives intoa coherent whole; to be able to impose losses on powerful groups; to representdiffuse,unorganized nterestsin addition to concentrated,well-organizedones; toensure effective implementation of government policies once they have beendecidedupon;to ensurepolicy stabilityso thatpolicies have time to work;to makeand maintain international commitments in the realms of trade and national defenseto ensuretheirlong-termwell-being; and,aboveall, to managepolitical cleavagesto ensure that society does not degenerateinto civil war.23

While Weaver andRockmanare interestedin thecapabilities of governments,a great volume of economic literature is concerned with the crediblecommitment of the government not to interfere with the economy (startingwithKydland and Prescott).24Weingast pushes the argument one step furtherandattemptsto design institutions thatwould produce such a credible commitment.His suggestion is that 'market preserving federalism' combines checks andbalances thatprevent government interference in the economy, with economiccompetition among units to assure growth.25In all these very diverse bodies of literaturethe flexibility or the stability ofpolicy is considered an importantvariable. Some scholars consider flexibilitya desirable feature (in order to resolve problems faster), whereas others pointout thatfrequentinterventions may worsen the situation. I take a more agnosticposition with respect to policy stability. Itseems reasonableto assume that thosewho dislike the statusquo will preferapolitical system with thecapacity to makechanges quickly, while advocates of the status quo will prefer a system thatproduces policy stability. Even if majorities are large (in which case theargument can be made that outcomes should conform to the will of thesemajorities), it may still be the case that institutional structureswill respond ata faster or slower pace thandesired. It is not clear to me that a consensus exists(or is even possible) over whether a faster or slower pace of institutionalresponse is desirable. Decisiveness in changing the status quo is good whenthe status quo is undesirable (whether it is because a small minority controlsthe government as in the French ancien regime or in South Africa recently), orwhen an exogenous shock disturbs a desirable process. Commitment to non-interference may be preferable when the status quo is desirable (as when civilrights are established), or if an exogenous shock is beneficial (like an increase

23R. Kent Weaver and Bert Rockman, Do InstitutionsMatter? GovermentCapabilities in the USand Abroad (Washington, DC: The Brookings Institution, 1993), p. 6.24Finn E. Kydlandand EdwardC. Prescott, 'Rules RatherthanDiscretion: The Inconsistency ofOptimal Plans', Journal of Political Economy, 85 (1977), 473-91.25BarryWeingast, 'Economic Role of Political Institutions' (mimeo, Hoover Institution, 1993).

example, in a thoughtful analysis of the effects of political institutions, Weaverand Rockman distinguish:ten differentcapabilitiesthatall governmentsneed: to set and maintainprioritiesamong the many conflicting demands made upon them so that they are notoverwhelmed and bankrupted;o target resources where they are most effective;to innovate when old policies have failed; to coordinate conflicting objectives intoa coherent whole; to be able to impose losses on powerful groups; to representdiffuse,unorganized nterestsin addition to concentrated,well-organizedones; toensure effective implementation of government policies once they have beendecidedupon;to ensurepolicy stabilityso thatpolicies have time to work;to makeand maintain international commitments in the realms of trade and national defenseto ensuretheirlong-termwell-being; and,aboveall, to managepolitical cleavagesto ensure that society does not degenerateinto civil war.23

While Weaver andRockmanare interestedin thecapabilities of governments,a great volume of economic literature is concerned with the crediblecommitment of the government not to interfere with the economy (startingwithKydland and Prescott).24Weingast pushes the argument one step furtherandattemptsto design institutions thatwould produce such a credible commitment.His suggestion is that 'market preserving federalism' combines checks andbalances thatprevent government interference in the economy, with economiccompetition among units to assure growth.25In all these very diverse bodies of literaturethe flexibility or the stability ofpolicy is considered an importantvariable. Some scholars consider flexibilitya desirable feature (in order to resolve problems faster), whereas others pointout thatfrequentinterventions may worsen the situation. I take a more agnosticposition with respect to policy stability. Itseems reasonableto assume that thosewho dislike the statusquo will preferapolitical system with thecapacity to makechanges quickly, while advocates of the status quo will prefer a system thatproduces policy stability. Even if majorities are large (in which case theargument can be made that outcomes should conform to the will of thesemajorities), it may still be the case that institutional structureswill respond ata faster or slower pace thandesired. It is not clear to me that a consensus exists(or is even possible) over whether a faster or slower pace of institutionalresponse is desirable. Decisiveness in changing the status quo is good whenthe status quo is undesirable (whether it is because a small minority controlsthe government as in the French ancien regime or in South Africa recently), orwhen an exogenous shock disturbs a desirable process. Commitment to non-interference may be preferable when the status quo is desirable (as when civilrights are established), or if an exogenous shock is beneficial (like an increase

23R. Kent Weaver and Bert Rockman, Do InstitutionsMatter? GovermentCapabilities in the USand Abroad (Washington, DC: The Brookings Institution, 1993), p. 6.24Finn E. Kydlandand EdwardC. Prescott, 'Rules RatherthanDiscretion: The Inconsistency ofOptimal Plans', Journal of Political Economy, 85 (1977), 473-91.25BarryWeingast, 'Economic Role of Political Institutions' (mimeo, Hoover Institution, 1993).


8/38

Decision Making in Political Systemsecision Making in Political Systemsin the oil price in an oil-producing economy). Finally, decisiveness is requiredwhen the default solution is not the previous status quo, but some otherdetrimentaldefault solution. In the restof the articleI shall speakaboutthe statusquo, but the more general term 'default solution' is more appropriate.Let menow turn to the concepts I shall use in the remainder of the article.Figure 1 is intended to familiarize the readerwith the concepts I use in therest of this analysis. Imagine a unicameral legislature making decisions overissues in two dimensions with three legislators (or threeparties), none of whomhas anabsolutemajority.Imaginefurther hat these threelegislators aredecidingthe size of the defence and social security budgets. Each legislator has an 'idealpoint' in the policy space, thatis, a most preferredcombination of budget sizes.In addition, each legislator is indifferent among budgets that are an equaldistance from his ideal point, in other words, each has a circular indifferencecurve.26In this case, if the legislature decides by a majorityof its members, andif its members are not on the same straightline, no matterwhere the status quois located, it can be defeated. Indeed, the three shaded petals in Figure 1 arecomposed of all points that can defeat the statusquo. I will call this shaded areathe winset of the status quo.I use the size of the winset of the status quo as a proxy for stability. Thereareseveral reasons for this. First,the more points (i.e. policy proposals) that candefeat the statusquo, the more susceptible to change is the statusquo. Secondly,the bigger the winset of the status quo is, the more likely it is that some subsetof it will satisfy some externalconstraints.Thirdly, if there aretransactioncostsin changing the status quo, then players will not undertake a change that leadsto a policy which is only slightly different, which means that the statusquo willremain.Fourthly,even without transactioncosts, if players undertakea change,a small winset of the status quo means that the change will be incremental. Inother words, a small winset of the status quo precludes majorpolicy changes.Each of these reasons is sufficient to justify the use of the size of the winset ofthe status quo as a proxy for policy stability. Consequently, I now turnto thevariables that affect the size of the winset of the status quo.Consider decision making that is delegated to one individual player (adictator,a charismaticleaderin a one-partysystem, or the leader of a disciplinedparty). By definition, the policies selected by this person will reflect his idealpoint. For this reason, the statusquo will follow the positions of this player, aslong as he remains the decision maker. If his preferredpolicy shifts from onepoint to another, the status quo will follow; and if he is replaced by anotherdecision maker, the status quo will move to her ideal point.If instead of a single decision maker there were two, they would prefer anypoint inside the intersection of their indifference curves over the statusquo. For

-' A morerealisticrepresentationwould havethe legislatorcare whethera budgetis above orbelowhis own point, as well as about other factors. While such complications would affect the simplicityof the presentationof the argumentthey would not affect its logic. I will proceed with the simplestexpositional convention of 'Euclidean preferences', that is, circular indifference curves.

in the oil price in an oil-producing economy). Finally, decisiveness is requiredwhen the default solution is not the previous status quo, but some otherdetrimentaldefault solution. In the restof the articleI shall speakaboutthe statusquo, but the more general term 'default solution' is more appropriate.Let menow turn to the concepts I shall use in the remainder of the article.Figure 1 is intended to familiarize the readerwith the concepts I use in therest of this analysis. Imagine a unicameral legislature making decisions overissues in two dimensions with three legislators (or threeparties), none of whomhas anabsolutemajority.Imaginefurther hat these threelegislators aredecidingthe size of the defence and social security budgets. Each legislator has an 'idealpoint' in the policy space, thatis, a most preferredcombination of budget sizes.In addition, each legislator is indifferent among budgets that are an equaldistance from his ideal point, in other words, each has a circular indifferencecurve.26In this case, if the legislature decides by a majorityof its members, andif its members are not on the same straightline, no matterwhere the status quois located, it can be defeated. Indeed, the three shaded petals in Figure 1 arecomposed of all points that can defeat the statusquo. I will call this shaded areathe winset of the status quo.I use the size of the winset of the status quo as a proxy for stability. Thereareseveral reasons for this. First,the more points (i.e. policy proposals) that candefeat the statusquo, the more susceptible to change is the statusquo. Secondly,the bigger the winset of the status quo is, the more likely it is that some subsetof it will satisfy some externalconstraints.Thirdly, if there aretransactioncostsin changing the status quo, then players will not undertake a change that leadsto a policy which is only slightly different, which means that the statusquo willremain.Fourthly,even without transactioncosts, if players undertakea change,a small winset of the status quo means that the change will be incremental. Inother words, a small winset of the status quo precludes majorpolicy changes.Each of these reasons is sufficient to justify the use of the size of the winset ofthe status quo as a proxy for policy stability. Consequently, I now turnto thevariables that affect the size of the winset of the status quo.Consider decision making that is delegated to one individual player (adictator,a charismaticleaderin a one-partysystem, or the leader of a disciplinedparty). By definition, the policies selected by this person will reflect his idealpoint. For this reason, the statusquo will follow the positions of this player, aslong as he remains the decision maker. If his preferredpolicy shifts from onepoint to another, the status quo will follow; and if he is replaced by anotherdecision maker, the status quo will move to her ideal point.If instead of a single decision maker there were two, they would prefer anypoint inside the intersection of their indifference curves over the statusquo. For

-' A morerealisticrepresentationwould havethe legislatorcare whethera budgetis above orbelowhis own point, as well as about other factors. While such complications would affect the simplicityof the presentationof the argumentthey would not affect its logic. I will proceed with the simplestexpositional convention of 'Euclidean preferences', that is, circular indifference curves.

29595


9/38

296 TSEBELIS96 TSEBELISIssue 2 \

SQ

Issue 1Fig. 1. Winsetof status quo with three players in two dimensions

example, in Figure 1 players A and C would prefer anything inside the shadedarea that represents the intersection of their indifference curves. Note that thisarea is a subset of the circle aroundA, that is, that the introductionof a secondplayer restricts the area of feasible outcomes.Let us increase the number of requirements now, and assume that theunanimousagreementof three individual players is requiredfor a change of thestatus quo. Consider the players A, B and C in Figure 2 and the status quo SQ.As long as the players remain in the same positions, the status quo cannot bechanged (since any change will be opposed by at least one of the players).Consider now thatplayerA changes his position fromA to A2. In this case, thestatusquo remains unchangedbecause players B and C are not willing to moveanywhere outside the areaof WBC,and playerA2would not like to move insidethis area. Despite the change of policy positions by playerA, there is no changein policy. On the other hand, if player A is replaced by player D, then changesin policy become possible. Indeed, any point inside WBCDcan defeat the statusquo and be selected by all three players.

Issue 2 \

SQ

Issue 1Fig. 1. Winsetof status quo with three players in two dimensions

example, in Figure 1 players A and C would prefer anything inside the shadedarea that represents the intersection of their indifference curves. Note that thisarea is a subset of the circle aroundA, that is, that the introductionof a secondplayer restricts the area of feasible outcomes.Let us increase the number of requirements now, and assume that theunanimousagreementof three individual players is requiredfor a change of thestatus quo. Consider the players A, B and C in Figure 2 and the status quo SQ.As long as the players remain in the same positions, the status quo cannot bechanged (since any change will be opposed by at least one of the players).Consider now thatplayerA changes his position fromA to A2. In this case, thestatusquo remains unchangedbecause players B and C are not willing to moveanywhere outside the areaof WBC,and playerA2would not like to move insidethis area. Despite the change of policy positions by playerA, there is no changein policy. On the other hand, if player A is replaced by player D, then changesin policy become possible. Indeed, any point inside WBCDcan defeat the statusquo and be selected by all three players.


10/38

Decision Making in Political Systemsecision Making in Political SystemsIssue2A

A1 A2.

/ //^-- WBCD

B C//Y

WBC

Issue 1Statusquocannotbe changedeven ifA1moves to A2.IfA1 s replacedbyD,thenthestatusquocan be replacedbyany point n the WBCD rea.

Fig. 2. Change of status quo with three individual decision makers

Some of these resultsdependon the numberof policy dimensions considered.For presentational purposes I use a two-dimensional space. For example, inFigure 2 we found that a movement of player A from A to A2did not have anyeffect on the statusquo. This result would not have been the same in more thantwo dimensions. However, there is one important conclusion that holds trueregardlessof the number of dimensions. I presentthis as a propositionto be usedthroughout the remainderof the article.

PROPOSITION 1: As the numberof players who are requiredto agree fora movement of the status quo increases, the winset of the status quo doesnot increase (i.e., policy stability does not decrease).The argumentbehind Proposition 1 is simple: the winset of the statusquo ofn + 1 players is a subset of the winset of the status quo of n players. For thisreason,addingone ormore veto playerswill never increase the size of thewinsetof the status quo.Consider now two individual players having to agree on a movement of thestatusquo SQ, as in Figure 3. If player B is close to playerA (position B1in thefigure), the winset of the status quo is WAB1.If, however, player B is further

Issue2A

A1 A2.

/ //^-- WBCD

B C//Y

WBC

Issue 1Statusquocannotbe changedeven ifA1moves to A2.IfA1 s replacedbyD,thenthestatusquocan be replacedbyany point n the WBCD rea.

Fig. 2. Change of status quo with three individual decision makers

Some of these resultsdependon the numberof policy dimensions considered.For presentational purposes I use a two-dimensional space. For example, inFigure 2 we found that a movement of player A from A to A2did not have anyeffect on the statusquo. This result would not have been the same in more thantwo dimensions. However, there is one important conclusion that holds trueregardlessof the number of dimensions. I presentthis as a propositionto be usedthroughout the remainderof the article.

PROPOSITION 1: As the numberof players who are requiredto agree fora movement of the status quo increases, the winset of the status quo doesnot increase (i.e., policy stability does not decrease).The argumentbehind Proposition 1 is simple: the winset of the statusquo ofn + 1 players is a subset of the winset of the status quo of n players. For thisreason,addingone ormore veto playerswill never increase the size of thewinsetof the status quo.Consider now two individual players having to agree on a movement of thestatusquo SQ, as in Figure 3. If player B is close to playerA (position B1in thefigure), the winset of the status quo is WAB1.If, however, player B is further

29797


11/38

298 TSEBELIS98 TSEBELIS

Issue 2

WAB1SQ

WAB

pPi2 B1 PB2B2

Issue 1WAB2 WAB1iffB1is between B2 and A

Fig. 3. Change of status quo as a function of the distance of individual decision makers

away from A on the same line the size of the winset of the statusquo is reduced.As we can see, WAB2s a subset of WAB,.We know this to be true because theside B2SQof the triangle BIB2SQ is smaller thanthe sum of the other two sides,and consequently, the distance B2P2 is smaller than the distance B2P1. This isanother general property which I shall single out for subsequent use.PROPOSITION2: As the distance of players who arerequiredto agree fora movement of the status quo increases along the same line, the winset ofthe status quo does not increase (i.e., policy stability increases).

Intheprevious examples, playerswere considered single individuals (orsomeotherentity that could reasonablybe assimilated to anindividual). Whathappensif players are collections of individuals without identical positions? I now turnto this point. I will assume collective players are composed of individuals with

Issue 2

WAB1SQ

WAB

pPi2 B1 PB2B2

Issue 1WAB2 WAB1iffB1is between B2 and A

Fig. 3. Change of status quo as a function of the distance of individual decision makers

away from A on the same line the size of the winset of the statusquo is reduced.As we can see, WAB2s a subset of WAB,.We know this to be true because theside B2SQof the triangle BIB2SQ is smaller thanthe sum of the other two sides,and consequently, the distance B2P2 is smaller than the distance B2P1. This isanother general property which I shall single out for subsequent use.PROPOSITION2: As the distance of players who arerequiredto agree fora movement of the status quo increases along the same line, the winset ofthe status quo does not increase (i.e., policy stability increases).

Intheprevious examples, playerswere considered single individuals (orsomeotherentity that could reasonablybe assimilated to anindividual). Whathappensif players are collections of individuals without identical positions? I now turnto this point. I will assume collective players are composed of individuals with


12/38

Decision Making in Political Systemsecision Making in Political Systemscircular indifference curves who decide by simple majorityrule. This discussionwill permit us to transpose the findings of the previous discussion to morerealistic situations with collective, rather than individual players.Social choice theoryhas demonstrated thatwithin every collective actorthereis a centrally located sphere which is called the 'yolk'.27The size r of the radiusof the yolk is usually very small, and on the average it decreases with the numberof individual voters with distinct positions.28If one calls C the centre of the yolkof a collective actor and d the distance of the statusquo (SQ) from C, the winsetof SQ for this actor is included in a sphere of centre C and radius d + 2r. Thisis an importantsocial choice finding for our purposes because it permits us toreplace the individual players in the previous figures with collective players.

circular indifference curves who decide by simple majorityrule. This discussionwill permit us to transpose the findings of the previous discussion to morerealistic situations with collective, rather than individual players.Social choice theoryhas demonstrated thatwithin every collective actorthereis a centrally located sphere which is called the 'yolk'.27The size r of the radiusof the yolk is usually very small, and on the average it decreases with the numberof individual voters with distinct positions.28If one calls C the centre of the yolkof a collective actor and d the distance of the statusquo (SQ) from C, the winsetof SQ for this actor is included in a sphere of centre C and radius d + 2r. Thisis an importantsocial choice finding for our purposes because it permits us toreplace the individual players in the previous figures with collective players.

Issue2AIssue2ARadius of yolk = rC, SQ=dRadius of yolk = rC, SQ=d

Issue 1ssue 1W (SQ) is included in a circle with centre at the centre of the yolk and radius d+ 2r

Fig. 4. Yolkand winset of SQ of a collective decision maker

27The yolk is defined as the smallest sphere that intersects all median hyperplanes. Hyperplanesare planes in more than two dimensions. A median hyperplane is a hyperplane that divides theindividual voters into three groups so that those voters on the hyperplaneor on one side of it canform a majority, as can those on it or on the other side of it. For a more complete discussion, seeJohn A. Ferejohn, Richard D. McKelvey and Edward W. Packell, 'Limiting Distributions forContinuous State Markov Voting Models', Social Choice and Welfare, 1 (1984), 45-67. For anon-technical discussion of the yolk and the calculation of winsets, see Nicholas R. Miller, BernardGrofman and Scott L. Feld, 'The Geometry of Majority Rule', Journal of Theoretical Politics, 4(1989), 379-406.28D. H. Koehler, 'The Size of the Yolk: ComputationsforOdd andEven-NumberedCommittees',Social Choice and Welfare, 7 (1990), 231-45.

W (SQ) is included in a circle with centre at the centre of the yolk and radius d+ 2r

Fig. 4. Yolkand winset of SQ of a collective decision maker

27The yolk is defined as the smallest sphere that intersects all median hyperplanes. Hyperplanesare planes in more than two dimensions. A median hyperplane is a hyperplane that divides theindividual voters into three groups so that those voters on the hyperplaneor on one side of it canform a majority, as can those on it or on the other side of it. For a more complete discussion, seeJohn A. Ferejohn, Richard D. McKelvey and Edward W. Packell, 'Limiting Distributions forContinuous State Markov Voting Models', Social Choice and Welfare, 1 (1984), 45-67. For anon-technical discussion of the yolk and the calculation of winsets, see Nicholas R. Miller, BernardGrofman and Scott L. Feld, 'The Geometry of Majority Rule', Journal of Theoretical Politics, 4(1989), 379-406.28D. H. Koehler, 'The Size of the Yolk: ComputationsforOdd andEven-NumberedCommittees',Social Choice and Welfare, 7 (1990), 231-45.

29999


13/38

300 TSEBELIS00 TSEBELISFigure 4 provides a visual representation of the argument. Five individualplayers form a collective actorwhose agreementby majorityrule is requiredfor

a change in the status quo. The figure indicates the yolk (centre C and radiusr) of this collective actor, and the winset of the status quo. It is easy to verifythat the winset of the statusquo is included in the circle with centre equal to thecentre of the yolk and radius d + 2r, where d is the distance between the statusquo and the centre C of the yolk.

Figure 4 provides a visual representation of the argument. Five individualplayers form a collective actorwhose agreementby majorityrule is requiredfora change in the status quo. The figure indicates the yolk (centre C and radiusr) of this collective actor, and the winset of the status quo. It is easy to verifythat the winset of the statusquo is included in the circle with centre equal to thecentre of the yolk and radius d + 2r, where d is the distance between the statusquo and the centre C of the yolk.

Issue 2A

Issue 2A

2rBrB

WABCABCA C

Issue 1ssue 1Statusquocannotbe replaced fdecisionmakersare individuals;t can be movedanywhere nsideWABCfdecisionmakersarecollective;rA, B, care the radiiof theyolksof playersA, B, C.

Fig. 5. Differences between individual and collective decision makers or the change of the statusquo; agreement of three players requiredfor decision

Figure 5 uses the argument presented in Figure 4 to replace the individualplayers with collective players. One can think of Figure 5 as the extension ofFigure 2 for the case of collective, rather than individual players. I call rA, rB,and rc the radii of the yolks of the collective players A, B and C respectively.In this case, the winset of the status quo includes points that are at greaterdistance from the centres of the yolks of the collective players than the statusquo itself. I have drawn the corresponding circles in Figure 5, and the set

Statusquocannotbe replaced fdecisionmakersare individuals;t can be movedanywhere nsideWABCfdecisionmakersarecollective;rA, B, care the radiiof theyolksof playersA, B, C.Fig. 5. Differences between individual and collective decision makers or the change of the statusquo; agreement of three players requiredfor decision

Figure 5 uses the argument presented in Figure 4 to replace the individualplayers with collective players. One can think of Figure 5 as the extension ofFigure 2 for the case of collective, rather than individual players. I call rA, rB,and rc the radii of the yolks of the collective players A, B and C respectively.In this case, the winset of the status quo includes points that are at greaterdistance from the centres of the yolks of the collective players than the statusquo itself. I have drawn the corresponding circles in Figure 5, and the set


14/38

Decision Making in Political Systemsecision Making in Political Systemsof points that can defeat the status quo is included inside WABC.29The nextproposition follows straightforwardly.

PROPOSITION3: As the size of the yolk of collective players who arerequiredto agree for a movement of the statusquo increases, the areathatincludes the winset of the status quo increases (i.e., policy stabilitydecreases).It is easy to see that individual players are merely a special case of collectiveplayers with yolk radiusequal to zero. However, as we will see below, therearealso collective players with a yolk size equal to zero who can be assimilated toindividual players for our purposes.To recapitulate, the size of the winset is a proxy for the policy stability of apolitical system. Propositions 1,2 and 3 give the size of the winset as a functionof different variables that we will consider subsequently.

II. THE INDEPENDENT VARIABLES: VETO PLAYERS AND THEIRCHARACTERISTICSIn this section I define the fundamentalconcept for understandingthe logic ofdifferent institutional settings: the veto player. I use Propositions 1, 2 and 3 todemonstrate that the policy stability of a political system depends on threecharacteristics of its veto players:theirnumber,theircongruence (thedifferencein theirpolitical positions) and theircohesion (the similarity of policy positionsof the constituent units of each veto player). While I use the number of vetoplayers, their congruence and cohesion as independentvariables, I discuss theconnections that have been made in the literaturebetween these variables andother institutional factors, mainly the electoral system.A veto player is an individual or collective actor whose agreement (bymajorityrule for collective actors) is requiredfor a change in policy. The vetoplayer concept stems from the idea of 'checks and balances' in the AmericanConstitution and the classic constitutional texts of the eighteenth century andlater, and is repeated implicitly or explicitly in contemporary studies.30Forexample, Montesquieu defends the need for bicameralism by arguing:In a state there arealways some people who aredistinguishedby birth,wealthorhonours,but if they were mixed amongthe people andif they hadonly one voicelike theothers,thecommonlibertywould be theirenslavementandtheywouldhaveno interest n defendingit, becausemost of the resolutionswould be againstthem.

29 In fact, one can locate the winset of the statusquo in a smaller area,but while such an increasein precision would greatly complicate the exposition it would not alterthe resultsreportedhere. Forsuch an example, see George Tsebelis, 'The Core, the Uncovered Set and Conference Committeesin Bicameral Legislatures' (paper presented at Annual Meeting of the American Political ScienceAssociation, 1993).30 See Lijphart,Democracies; Riker, 'The Merits of Bicameralism'; and Weaver and Rockman,Do InstitutionsMatter?

of points that can defeat the status quo is included inside WABC.29The nextproposition follows straightforwardly.PROPOSITION3: As the size of the yolk of collective players who arerequiredto agree for a movement of the statusquo increases, the areathatincludes the winset of the status quo increases (i.e., policy stabilitydecreases).

It is easy to see that individual players are merely a special case of collectiveplayers with yolk radiusequal to zero. However, as we will see below, therearealso collective players with a yolk size equal to zero who can be assimilated toindividual players for our purposes.To recapitulate, the size of the winset is a proxy for the policy stability of apolitical system. Propositions 1,2 and 3 give the size of the winset as a functionof different variables that we will consider subsequently.

II. THE INDEPENDENT VARIABLES: VETO PLAYERS AND THEIRCHARACTERISTICSIn this section I define the fundamentalconcept for understandingthe logic ofdifferent institutional settings: the veto player. I use Propositions 1, 2 and 3 todemonstrate that the policy stability of a political system depends on threecharacteristics of its veto players:theirnumber,theircongruence (thedifferencein theirpolitical positions) and theircohesion (the similarity of policy positionsof the constituent units of each veto player). While I use the number of vetoplayers, their congruence and cohesion as independentvariables, I discuss theconnections that have been made in the literaturebetween these variables andother institutional factors, mainly the electoral system.A veto player is an individual or collective actor whose agreement (bymajorityrule for collective actors) is requiredfor a change in policy. The vetoplayer concept stems from the idea of 'checks and balances' in the AmericanConstitution and the classic constitutional texts of the eighteenth century andlater, and is repeated implicitly or explicitly in contemporary studies.30Forexample, Montesquieu defends the need for bicameralism by arguing:In a state there arealways some people who aredistinguishedby birth,wealthorhonours,but if they were mixed amongthe people andif they hadonly one voicelike theothers,thecommonlibertywould be theirenslavementandtheywouldhaveno interest n defendingit, becausemost of the resolutionswould be againstthem.

29 In fact, one can locate the winset of the statusquo in a smaller area,but while such an increasein precision would greatly complicate the exposition it would not alterthe resultsreportedhere. Forsuch an example, see George Tsebelis, 'The Core, the Uncovered Set and Conference Committeesin Bicameral Legislatures' (paper presented at Annual Meeting of the American Political ScienceAssociation, 1993).30 See Lijphart,Democracies; Riker, 'The Merits of Bicameralism'; and Weaver and Rockman,Do InstitutionsMatter?

30101


15/38

302 TSEBELIS02 TSEBELISTherefore,the part they have in legislation should be in proportionto the otheradvantages they have in the state, which will happenif they form a body that hasthe right to check the enterprises of the people, as the people have the right to checktheirs.31

Similarly, Madison, in Federalist No. 51, defends the separationof powersin the following way: 'contriving the interiorstructureof the government as thatits several constituent parts may, by their mutual relations, be the means ofkeeping each other in their proper places'.32 These texts both discuss vetoplayers specified by the Constitution. I shall call these institutional veto players.There is, however, anothercategory of veto players that exists in multi-partyparliamentary systems, and possibly in presidential systems as well: the partiesthat are members of a government coalition. I will call these coalition memberspartisan veto players.To simplify matters,I assume that agovernment proposal has to be approvedby a majority of the relevant actors within each party of the governmentcoalition.33 This is only a first approximation. It assumes that there is nodifference between the approval of a policy by the majorities in the upperandlower chambers of a bicameral system (institutional veto players), and theapproval of a policy by the majorities of the two partners of a governmentcoalition (partisan veto players). Of course, there is one importantdifferencebetween institutional and partisan veto players: according to the constitution,the agreementof institutional veto players is a necessary andsufficient conditionfor policy change, while the agreement of partisan veto players is, strictlyspeaking, neither necessary nor sufficient.Agreement of partisanveto players is not sufficient forpolicy change becausea proposal which is approvedby all partnersin a government coalition may bedefeated in parliament, in which case no law is adopted. This is a case of anon-enforceable coalition agreement. The parties participating in governmentlack the resources to prevent their own MPs from defecting on parliamentaryvotes. Examples of such cases include the French FourthRepublic andpost-warItaly. It is ironic thatwhat General de Gaulle sneeringly referredto as the 'regimedes partis' was suffering precisely from the opposite: a lack of parties. Thedifferences were not over policies, but over personalities and the distributionofgovernment portfolios. Similarly, in Italy,franchi tiratori were taking advan-tage of the system of secret ballots to embarrass the government with defeats

31 Montesquieu, The Spirit of the Laws, A. M. Cohler, B. C. Miller, and H. S. Stone, trans. andeds (Cambridge:Cambridge University Press, 1989), part 2, bk. 11, chap. 6 (emphasis added).32 Alexander Hamilton, John Jay and James Madison, The Federalist (any edition: emphasisadded).33 Here is how Maorreportsthe position of a leader of the liberalparty,memberof the governmentcoalition in Denmark: 'We could stop everything we did not like. That is a problem with a coalition

governmentbetween two partiesof very differentprinciples. If you cannotreach a compromise, thensuch a government has to stay away from legislation in such areas.' See Moshe Maor, 'Intra-PartyConflict and Coalitional Behavior in Denmarkand Norway: The Case of Highly InstitutionalizedParties', Scandinavian Political Studies, 15 (1992), 99-116.

Therefore,the part they have in legislation should be in proportionto the otheradvantages they have in the state, which will happenif they form a body that hasthe right to check the enterprises of the people, as the people have the right to checktheirs.31Similarly, Madison, in Federalist No. 51, defends the separationof powersin the following way: 'contriving the interiorstructureof the government as thatits several constituent parts may, by their mutual relations, be the means of

keeping each other in their proper places'.32 These texts both discuss vetoplayers specified by the Constitution. I shall call these institutional veto players.There is, however, anothercategory of veto players that exists in multi-partyparliamentary systems, and possibly in presidential systems as well: the partiesthat are members of a government coalition. I will call these coalition memberspartisan veto players.To simplify matters,I assume that agovernment proposal has to be approvedby a majority of the relevant actors within each party of the governmentcoalition.33 This is only a first approximation. It assumes that there is nodifference between the approval of a policy by the majorities in the upperandlower chambers of a bicameral system (institutional veto players), and theapproval of a policy by the majorities of the two partners of a governmentcoalition (partisan veto players). Of course, there is one importantdifferencebetween institutional and partisan veto players: according to the constitution,the agreementof institutional veto players is a necessary andsufficient conditionfor policy change, while the agreement of partisan veto players is, strictlyspeaking, neither necessary nor sufficient.Agreement of partisanveto players is not sufficient forpolicy change becausea proposal which is approvedby all partnersin a government coalition may bedefeated in parliament, in which case no law is adopted. This is a case of anon-enforceable coalition agreement. The parties participating in governmentlack the resources to prevent their own MPs from defecting on parliamentaryvotes. Examples of such cases include the French FourthRepublic andpost-warItaly. It is ironic thatwhat General de Gaulle sneeringly referredto as the 'regimedes partis' was suffering precisely from the opposite: a lack of parties. Thedifferences were not over policies, but over personalities and the distributionofgovernment portfolios. Similarly, in Italy,franchi tiratori were taking advan-tage of the system of secret ballots to embarrass the government with defeats

31 Montesquieu, The Spirit of the Laws, A. M. Cohler, B. C. Miller, and H. S. Stone, trans. andeds (Cambridge:Cambridge University Press, 1989), part 2, bk. 11, chap. 6 (emphasis added).32 Alexander Hamilton, John Jay and James Madison, The Federalist (any edition: emphasisadded).33 Here is how Maorreportsthe position of a leader of the liberalparty,memberof the governmentcoalition in Denmark: 'We could stop everything we did not like. That is a problem with a coalition

governmentbetween two partiesof very differentprinciples. If you cannotreach a compromise, thensuch a government has to stay away from legislation in such areas.' See Moshe Maor, 'Intra-PartyConflict and Coalitional Behavior in Denmarkand Norway: The Case of Highly InstitutionalizedParties', Scandinavian Political Studies, 15 (1992), 99-116.


16/38

Decision Making in Political Systemsecision Making in Political Systemsin Parliament.34 In both of these cases, the stability of policy making increasesbecause the government is unable to modify the legislation that it intends toabolish. Below we see how we can take into account such increased stabilityinside the framework of veto players.

Agreement of partisan veto players is not necessary for policy change becausecoalition partners may be bypassed or played off against each other. There aretwo situations in which this can occur: minority governments, and oversizedmajority governments.

Minority governments may have their proposals approved by Parliament.Strom has analysed minority governments and finds that they are common inmulti-party systems (around one-third of the governments in his sample).35Moreover, most of them (79 out of 125) are single-party governments whichresemble single-party majority governments. Further, Laver and Schofield haveargued that there is a difference between a governmental and a legislativemajority, and that the party forming the minority government is usually locatedcentrally in space. For this reason, it can lean slightly towards one or anotherpossible partner in order to have its policies approved by parliament.36Consequently, from a policy-making point of view, a single-party minoritygovernment, as long as it stays in power, resembles a single-party majoritygovernment. There are two reasons for this. First, when the minority partyoccupies a centrally located area in space (technically speaking, the core), itneeds no formal allies, as Laver and Schofield, Strom and many others haveindicated.37 Secondly, regardless of the location of the party in government,several constitutions provide ruling governments with a series of agenda-settingpowers, such as the giving of priority to government bills, the possibility ofclosed or restricted rules, the practice of counting abstentions in favour ofgovernment bills, the possibility of introducing amendments at any point of thedebate (including before the final vote), and others. The most frequent andserious of all these agenda-setting measures is the threat of governmentresignation followed by the dissolution of parliament. This measure exists in allparliamentary systems with the exception of Norway.38Oversized majority governments are almost as common as minoritygovernments in Western Europe. Laver and Schofield calculate that 4 per centof the time (of the 218 governments they examine), a party which forms amajority alone will ask another party to join the government; and 21 per cent

34 The government introduced open votes in 1988 and did away with the problem.35 Strom, Minority Government and Majority Rule, p. 61.36 Michael J. Laver and Norman Schofield, MultipartyGovernment:The Politics of Coalition inEurope (Oxford: Oxford University Press, 1992).37 Laver and Schofield, MultipartyGovernment;and Strom, Minority Governmentand Majority

Rule. This idea originated with Anthony Downs, An Economic Theory of Democracy (New York:Harper& Row, 1957).38 For a systematic discussion of positional and institutional advantages of governments inparliamentarydemocracies, see George Tsebelis, 'Veto Playersand Law Production nParliamentaryDemocracies' in HerberDoering, ed. Parliaments andMajorityRule in WesternEurope(New York:St Martin's Press, forthcoming).

in Parliament.34 In both of these cases, the stability of policy making increasesbecause the government is unable to modify the legislation that it intends toabolish. Below we see how we can take into account such increased stabilityinside the framework of veto players.

Agreement of partisan veto players is not necessary for policy change becausecoalition partners may be bypassed or played off against each other. There aretwo situations in which this can occur: minority governments, and oversizedmajority governments.

Minority governments may have their proposals approved by Parliament.Strom has analysed minority governments and finds that they are common inmulti-party systems (around one-third of the governments in his sample).35Moreover, most of them (79 out of 125) are single-party governments whichresemble single-party majority governments. Further, Laver and Schofield haveargued that there is a difference between a governmental and a legislativemajority, and that the party forming the minority government is usually locatedcentrally in space. For this reason, it can lean slightly towards one or anotherpossible partner in order to have its policies approved by parliament.36Consequently, from a policy-making point of view, a single-party minoritygovernment, as long as it stays in power, resembles a single-party majoritygovernment. There are two reasons for this. First, when the minority partyoccupies a centrally located area in space (technically speaking, the core), itneeds no formal allies, as Laver and Schofield, Strom and many others haveindicated.37 Secondly, regardless of the location of the party in government,several constitutions provide ruling governments with a series of agenda-settingpowers, such as the giving of priority to government bills, the possibility ofclosed or restricted rules, the practice of counting abstentions in favour ofgovernment bills, the possibility of introducing amendments at any point of thedebate (including before the final vote), and others. The most frequent andserious of all these agenda-setting measures is the threat of governmentresignation followed by the dissolution of parliament. This measure exists in allparliamentary systems with the exception of Norway.38Oversized majority governments are almost as common as minoritygovernments in Western Europe. Laver and Schofield calculate that 4 per centof the time (of the 218 governments they examine), a party which forms amajority alone will ask another party to join the government; and 21 per cent

34 The government introduced open votes in 1988 and did away with the problem.35 Strom, Minority Government and Majority Rule, p. 61.36 Michael J. Laver and Norman Schofield, MultipartyGovernment:The Politics of Coalition inEurope (Oxford: Oxford University Press, 1992).37 Laver and Schofield, MultipartyGovernment;and Strom, Minority Governmentand Majority

Rule. This idea originated with Anthony Downs, An Economic Theory of Democracy (New York:Harper& Row, 1957).38 For a systematic discussion of positional and institutional advantages of governments inparliamentarydemocracies, see George Tsebelis, 'Veto Playersand Law Production nParliamentaryDemocracies' in HerberDoering, ed. Parliaments andMajorityRule in WesternEurope(New York:St Martin's Press, forthcoming).

30303


17/38

304 TSEBELIS04 TSEBELISof the time, while there is no majority party, the coalition formed contains oneor more parties more than necessary.39In these cases, some of the coalitionpartnerscan be disregarded, and policies will still be passed by a majority inparliament. Such a situation occurs frequently in Italy, where five partiesparticipatedin the governments of the 1980s. The ChristianDemocrats and theSocialists together had a majority of seats, making the other three partnersnumerically unnecessary. However, ignoring coalition partners,while numeric-ally possible, imposes political costs. If the disagreement is serious the smallpartner can resign, and the government formation process must begin overagain. Simple arithmeticdisregardsthe fact that therearepolitical factors whichnecessitate oversized coalitions. Regardless of what these reasons might be, forthe coalition to remainintact,the will of the differentpartnersmustbe respected.For this reason, each partnerin the coalition is a veto player. Consequently,while the arithmetic of the legislative process may be different from thearithmetic of government, a departure from the status quo must usually beapproved by the government before it is introduced to parliament and, at thatstage, the participants in a government coalition are veto players.In general, either constitutionally, or through the coalition bargainingprocess, the government is given extraordinary agenda-setting powers. Anexample of the former is the extraordinary legal arsenal of which the Frenchgovernment disposes (particularlyArticle 49.3 of the Constitution) allowing itto avoid amendments and even final votes on the floor of Parliament.40Anexample of the latter is the following statement from the Norwegian PrimeMinister Kare Willoch regarding his coalition government: 'I wanted theirleading personalities in the government. It was my demand that their partyleaders should be in government because I did not want to strengthenthe othercentres which would be in parliament. That was my absolute condition forhaving three parties in government'.41The outcome of all these agenda-setting procedures is that in more than 50per cent of all countries, governments introduce more than 90 per cent of thebills. Moreover, the probability of success for these bills is very high: over 60per cent pass with probability greaterthan 0.9, and over 85 per cent pass withprobability greater than 0.8.42In sum, one can say that whereas the number of institutional veto players isspecified by the Constitution, the number of partisanveto players is specified

39Laver and Schofield, MultipartyGovernment,p. 70.40 See John D. Huber, 'Restrictive Legislative Procedures in France and the US', AmericanPolitical Science Review, 86 (1992), 675-87; and George Tsebelis, Nested Games (Berkeley:University of California Press, 1990), chap. 7.41Quoted in Maor, 'Intra-PartyConflict and Coalitional Behavior in Denmark and Norway',p. 108.42 What these numbers do not specify, however, is how many amendmentswere made to the billsor, how many times the government may have altered the bill in anticipationof amendments. Fordata see Inter-ParliamentaryUnion, Parliaments of the World,2nd edn (Aldershot, Surrey:Gower,1986), Table 29.

of the time, while there is no majority party, the coalition formed contains oneor more parties more than necessary.39In these cases, some of the coalitionpartnerscan be disregarded, and policies will still be passed by a majority inparliament. Such a situation occurs frequently in Italy, where five partiesparticipatedin the governments of the 1980s. The ChristianDemocrats and theSocialists together had a majority of seats, making the other three partnersnumerically unnecessary. However, ignoring coalition partners,while numeric-ally possible, imposes political costs. If the disagreement is serious the smallpartner can resign, and the government formation process must begin overagain. Simple arithmeticdisregardsthe fact that therearepolitical factors whichnecessitate oversized coalitions. Regardless of what these reasons might be, forthe coalition to remainintact,the will of the differentpartnersmustbe respected.For this reason, each partnerin the coalition is a veto player. Consequently,while the arithmetic of the legislative process may be different from thearithmetic of government, a departure from the status quo must usually beapproved by the government before it is introduced to parliament and, at thatstage, the participants in a government coalition are veto players.In general, either constitutionally, or through the coalition bargainingprocess, the government is given extraordinary agenda-setting powers. Anexample of the former is the extraordinary legal arsenal of which the Frenchgovernment disposes (particularlyArticle 49.3 of the Constitution) allowing itto avoid amendments and even final votes on the floor of Parliament.40Anexample of the latter is the following statement from the Norwegian PrimeMinister Kare Willoch regarding his coalition government: 'I wanted theirleading personalities in the government. It was my demand that their partyleaders should be in government because I did not want to strengthenthe othercentres which would be in parliament. That was my absolute condition forhaving three parties in government'.41The outcome of all these agenda-setting procedures is that in more than 50per cent of all countries, governments introduce more than 90 per cent of thebills. Moreover, the probability of success for these bills is very high: over 60per cent pass with probability greaterthan 0.9, and over 85 per cent pass withprobability greater than 0.8.42In sum, one can say that whereas the number of institutional veto players isspecified by the Constitution, the number of partisanveto players is specified

39Laver and Schofield, MultipartyGovernment,p. 70.40 See John D. Huber, 'Restrictive Legislative Procedures in France and the US', AmericanPolitical Science Review, 86 (1992), 675-87; and George Tsebelis, Nested Games (Berkeley:University of California Press, 1990), chap. 7.41Quoted in Maor, 'Intra-PartyConflict and Coalitional Behavior in Denmark and Norway',p. 108.42 What these numbers do not specify, however, is how many amendmentswere made to the billsor, how many times the government may have altered the bill in anticipationof amendments. Fordata see Inter-ParliamentaryUnion, Parliaments of the World,2nd edn (Aldershot, Surrey:Gower,1986), Table 29.


18/38

Decision Making in Political Systemsecision Making in Political Systemsendogenously by the party system and the government coalitions of eachspecific country. However, both categories of veto players are easilyidentifiable. In this article I shall restrict my discussion to these two groups ofveto players, but I shall also indicate later how the logic of the analysis caninclude other veto players, and how such players can be identified.Once institutional andpartisanveto players in a system areidentified, we canuse Propositions 1, 2 and 3 to calculate the size of the winset of the statusquo,and therefore the policy stability of that system. The remainderof this sectionis a discussion of the political aspects of Propositions 1, 2 and 3.

Number of Veto PlayersFromthe definition of veto players, thecounting rules arestraightforward:a vetoplayer is any player - institutionalor partisan- who can block the adoption ofa policy. Since, however, this counting rule sometimes results in outcomes thatgo against received wisdom, I shall be more specific about how to count.An institutionalplayerwill not count as a veto playerunless it has formal vetopower. With respect to bicameralism, there are countries where the upperchamber has only a delaying veto power.43For example, while Britain andAustria areformally bicameral systems, in both systems the lower chamber canultimately overrule the objections of the upper chamber. Consequently, thesetwo systems must be classified as unicameral legislatures. A complete list ofbicameral legislatures and their decision-making rules can be found in theliterature, so I will not duplicate the information here.44 However, someimportant cases should be mentioned. For example, for counting purposesFrance is a unicamerallegislature. Germany,on the otherhand, is a mixed case:only legislation concerningfederalism requiresthe agreementof bothchambers.But the number of such laws requiring the agreement of the BundesratZustimmungsgesetze,i.e. agreement laws) has increased over time to more than50 per cent. Consequently, depending on the issue, the numberof institutionalveto players in Germany is either one or two.With respectto presidentialism,not all popularlyelected presidentshave vetopowers, and when they do, their veto can almost always be overruled by anappropriatemajority in the legislature.45I will not discuss the complication ofveto override here.46The point I want to make is that in several presidentialregimes - Venezuela, Haiti and Peru - the presidentdoes not have veto powers

43 Lijphartcalls these bicamerallegislatures asymmetric.See Lijphart,Democracies, pp. 95-100.44Money and Tsebelis, 'Cicero's Puzzle'.45 With the exception of the PortugueseConstitution of 1976, which was revised on this and other

points in 1982. See Shugartand Carey, Presidents and Assemblies, p. 155.46 Fora discussion of veto override,see Hammond andMiller, 'The Core of theConstitution'. Theessence of the argumentis that if players House (H), Senate (S), and President(P) have veto powersbut P's veto can be overruled,then the finaloutcome canbe notonly in the intersection of the winsetsof H, S, and P, but also in some partof the intersection of the winsets of H and S which does notbelong to the winset of P.

endogenously by the party system and the government coalitions of eachspecific country. However, both categories of veto players are easilyidentifiable. In this article I shall restrict my discussion to these two groups ofveto players, but I shall also indicate later how the logic of the analysis caninclude other veto players, and how such players can be identified.Once institutional andpartisanveto players in a system areidentified, we canuse Propositions 1, 2 and 3 to calculate the size of the winset of the statusquo,and therefore the policy stability of that system. The remainderof this sectionis a discussion of the political aspects of Propositions 1, 2 and 3.

Number of Veto PlayersFromthe definition of veto players, thecounting rules arestraightforward:a vetoplayer is any player - institutionalor partisan- who can block the adoption ofa policy. Since, however, this counting rule sometimes results in outcomes thatgo against received wisdom, I shall be more specific about how to count.An institutionalplayerwill not count as a veto playerunless it has formal vetopower. With respect to bicameralism, there are countries where the upperchamber has only a delaying veto power.43For example, while Britain andAustria areformally bicameral systems, in both systems the lower chamber canultimately overrule the objections of the upper chamber. Consequently, thesetwo systems must be classified as unicameral legislatures. A complete list ofbicameral legislatures and their decision-making rules can be found in theliterature, so I will not duplicate the information here.44 However, someimportant cases should be mentioned. For example, for counting purposesFrance is a unicamerallegislature. Germany,on the otherhand, is a mixed case:only legislation concerningfederalism requiresthe agreementof bothchambers.But the number of such laws requiring the agreement of the BundesratZustimmungsgesetze,i.e. agreement laws) has increased over time to more than50 per cent. Consequently, depending on the issue, the numberof institutionalveto players in Germany is either one or two.With respectto presidentialism,not all popularlyelected presidentshave vetopowers, and when they do, their veto can almost always be overruled by anappropriatemajority in the legislature.45I will not discuss the complication ofveto override here.46The point I want to make is that in several presidentialregimes - Venezuela, Haiti and Peru - the presidentdoes not have veto powers

43 Lijphartcalls these bicamerallegislatures asymmetric.See Lijphart,Democracies, pp. 95-100.44Money and Tsebelis, 'Cicero's Puzzle'.45 With the exception of the PortugueseConstitution of 1976, which was revised on this and other

points in 1982. See Shugartand Carey, Presidents and Assemblies, p. 155.46 Fora discussion of veto override,see Hammond andMiller, 'The Core of theConstitution'. Theessence of the argumentis that if players House (H), Senate (S), and President(P) have veto powersbut P's veto can be overruled,then the finaloutcome canbe notonly in the intersection of the winsetsof H, S, and P, but also in some partof the intersection of the winsets of H and S which does notbelong to the winset of P.

30505


19/38

306 TSEBELIS06 TSEBELISand so does not count as a veto player. In addition, in all the regimes thatDuverger labels semi-presidential (with theexception of Portugal),thepresidentdoes not have veto powers.47 Consequently, a group of popularly electedpresidents, even some of them who areconsidered 'strong' - like the presidentsof France,Finlandand the WeimarRepublic - along with the weaker popularlyelected presidentsof Ireland,Iceland andAustria,are not veto players accordingto my definition.48 A complete list of presidential powers in different countriesis provided by Shugart and Carey and will not be duplicated here.)49At this point the reader may object that my argument does not produceunambiguous classifications - for instance, Germany is sometimes unicameraland sometimes bicameral in terms of institutional actors - and seems tomisclassify countries - for example, France with a strong president and abicameral legislature is considered a pure parliamentary regime with oneinstitutional veto player. My response to such objections is to plead guilty andclaim that it is the constitutional requirements and the logic of policy makingwithin these countries that lead to these results. The traditional classificationsof 'strong' or 'weak' presidents arebased not only on their legislative roles butalso on their non-legislative powers (to appoint or dismiss the government, toask for referendums, to declare a state of emergency), which are criteriairrelevant to my analysis.

Counting becomes more difficult when applied to partisanveto players. Thesame parliamentarysystem is classified as a one veto-player system when it hasa minority government, but as a two or three veto-player system when two orthree parties are in government. It is not only possible but also frequentthatthetransition from one government to another occurs without an election. In fact,this is a frequentcomplaint about parliamentarydemocracy: it is mediated, andthe link between the popular vote and the government coalition is not alwaysclear.There are two additional questions which need to be addressed with respectto counting: (1) Are institutional and partisanplayers the only veto players thatexist in a system? (2) How do we count in the presence of both institutionaland

partisan actors?The answer to the first question is negative. There are several additionalcategories of veto players in different political systems. For example, one canthink of powerful interest groups as veto players, at least in the policy areas oftheir concern. The army could also be a group of particular importance.50Political systems with few veto players may delegate policy making to several

47 MauriceDuverger, 'A New Political System Model: Semi-PresidentialGovernment', EuropeanJournal of Political Research, 8 (1980), 165-87.48 In fact thepresidentof theWeimarRepublic had anindirector conditionalveto: he could submitlegislation he did not like to a referendum.49 See Shugart and Carey, Presidents and Assemblies.50See BarryAmes, Political Survival in LatinAmerica (Berkeley: University of CaliforniaPress,1987).

and so does not count as a veto player. In addition, in all the regimes thatDuverger labels semi-presidential (with theexception of Portugal),thepresidentdoes not have veto powers.47 Consequently, a group of popularly electedpresidents, even some of them who areconsidered 'strong' - like the presidentsof France,Finlandand the WeimarRepublic - along with the weaker popularlyelected presidentsof Ireland,Iceland andAustria,are not veto players accordingto my definition.48 A complete list of presidential powers in different countriesis provided by Shugart and Carey and will not be duplicated here.)49At this point the reader may object that my argument does not produceunambiguous classifications - for instance, Germany is sometimes unicameraland sometimes bicameral in terms of institutional actors - and seems tomisclassify countries - for example, France with a strong president and abicameral legislature is considered a pure parliamentary regime with oneinstitutional veto player. My response to such objections is to plead guilty andclaim that it is the constitutional requirements and the logic of policy makingwithin these countries that lead to these results. The traditional classificationsof 'strong' or 'weak' presidents arebased not only on their legislative roles butalso on their non-legislative powers (to appoint or dismiss the government, toask for referendums, to declare a state of emergency), which are criteriairrelevant to my analysis.

Counting becomes more difficult when applied to partisanveto players. Thesame parliamentarysystem is classified as a one veto-player system when it hasa minority government, but as a two or three veto-player system when two orthree parties are in government. It is not only possible but also frequentthatthetransition from one government to another occurs without an election. In fact,this is a frequentcomplaint about parliamentarydemocracy: it is mediated, andthe link between the popular vote and the government coalition is not alwaysclear.There are two additional questions which need to be addressed with respectto counting: (1) Are institutional and partisanplayers the only veto players thatexist in a system? (2) How do we count in the presence of both institutionaland

partisan actors?The answer to the first question is negative. There are several additionalcategories of veto players in different political systems. For example, one canthink of powerful interest groups as veto players, at least in the policy areas oftheir concern. The army could also be a group of particular importance.50Political systems with few veto players may delegate policy making to several

47 MauriceDuverger, 'A New Political System Model: Semi-PresidentialGovernment', EuropeanJournal of Political Research, 8 (1980), 165-87.48 In fact thepresidentof theWeimarRepublic had anindirector conditionalveto: he could submitlegislation he did not like to a referendum.49 See Shugart and Carey,

George Tsebelis Veto Players

Documents

Transcript of George Tsebelis Veto Players