World Politics, Oct 1993 v46 n1 p1(22)
Constitutional frameworks and democratic consolidation: parliamentarianism versus presidentialism. Alfred Stepan; Cindy Skach.
COPYRIGHT 1993 Johns Hopkins University Press

A fundamental political-institutional question that has only recently received serious scholarly attention concerns the impact of different constitutional frameworks on democratic consolidation. Little systematic cross-regional evidence has been brought to bear on this question. This article reports the findings of the analysis of numerous different sources of data, all of which point in the direction of a much stronger correlation between democratic consolidation and the constitutional framework of pure parliamentarianism than between consolidation and pure presidentialism. The systematic analysis of these data leads the authors to conclude that parliamentarianism is a more supportive constitutional framework due to the following theoretically predictable and empirically observable tendencies: its greater propensity for governments to have majorities to implement their programs, its greater ability to rule in a multiparty setting, its lower propensity for executives to rule at the edge of the constitution and its greater facility in removing a chief executive if he or she does so, its lower susceptibility to a military coup, and its greater tendency to provide long party-government careers, which add loyalty and experience to political society. In contrast, the analytically separable propensities of presidentialism also form a highly interactive system, but they work to impede democratic consolidation by reducing politicians, degrees of freedom.

Full Text: COPYRIGHT 1993 Johns Hopkins University Press

Introduction

The struggle to consolidate the new democracies - especially those in Eastern Europe, Latin America, and Asia - has given rise to a wide-ranging debate about the hard choices concerning economic restructuring, economic institutions, and economic markets.(1) A similar debate has focused on democratic political institutions and political, markets. This literature has produced provocative hypotheses about the effects of institutions on democracy. It forms part of the "new institutionalism" literature in comparative politics that holds as a premise that "political democracy depends not only on economic and social conditions but also on the design of political institutions."(2)

One fundamental political-institutional question that has only recently received serious scholarly attention concerns the impact of different constitutional frameworks on democratic consolidation.(3) Although the topic has been increasingly debated and discussed, little systematic cross-regional evidence has been brought to bear on it. This is unfortunate, because constitutions are essentially "institutional frameworks" that in functioning democracies provide the basic decision rules and incentive systems concerning government formation, the conditions under which governments can continue to rule, and the conditions by which they can be terminated democratically. More than simply one of the many dimensions of a democratic system,(4) constitutions create much of the overall system of incentives and organizations within which the other institutions and dimensions found in the many types of democracy are structured and processed.

Study shows that the range of existing constitutional frameworks in the world's long-standing democracies is narrower than one would think.(5) With one exception (Switzerland), every existing democracy today is either presidential (as in the United States), parliamentary (as in most of Western Europe), or a semipresidential hybrid of the two (as in France and Portugal, where there is a directly elected president and a prime minister who must have a majority in the legislature).(6) In this essay we pay particular attention to contrasting what we call "pure parliamentarianism" with "pure presidentialism."(7) Each type has only two fundamental characteristics, and for our purposes of classification these characteristics are necessary and sufficient.

A pure parliamentary regime in a democracy is a system of mutual dependence:

1. The chief executive power must be supported by a majority in the legislature and can fall if it receives a vote of no confidence.

2. The executive power (normally in conjunction with the head of state) has the capacity to dissolve the legislature and call for elections.

A pure presidential regime in a democracy is a system of mutual independence:

1. The legislative power has a fixed electoral mandate that is its own source of legitimacy.

2. The chief executive power has a fixed electoral mandate that is its own source of legitimacy.

These necessary and sufficient characteristics are more than classificatory. They are also the constraining conditions within which the vast majority of aspiring democracies must somehow attempt simultaneously to produce major socioeconomic changes and to strengthen democratic institutions.(8)

Pure parliamentarianism, as defined here, had been the norm in the democratic world following World War II.(9) However, so far, in the 1980s and 1990s, all the new aspirant democracies in Latin America and Asia (Korea and the Philippines) have chosen pure presidentialism. And to date, of the approximately twenty-five countries that now constitute Eastern Europe and the former Soviet Union, only three - Hungary, the new Czech Republic, and Slovakia - have chosen pure parliamentarianism.(10)

We question the wisdom of this virtual dismissal of the pure parliamentary model by most new democracies and believe that the hasty embrace of presidential models should be reconsidered. In this article we bring evidence in support of the theoretical argument that parliamentary democracies tend to increase the degrees of freedom that facilitate the momentous tasks of economic and social restructuring facing new democracies as they simultaneously attempt to consolidate their democratic institutions.

It is not our purpose in this article to weigh the benefits and the draw-backs of parliamentarianism and presidentialism. Our intention is to report and analyze numerous different sources of data, all of which point in the direction of a much stronger correlation between democratic consolidation and pure parliamentarianism than between democratic consolidation and pure presidentialism. We believe our findings are sufficiently strong to warrant long-range studies that test the probabilistic propositions we indicate.(11)

Constitutional Frameworks: Constructing Relevant Data

We were able to construct a data set about party systems and consolidated democracies. Since we are interested in the lessons about party systems in long-standing consolidated democracies, we include the countries of the Organization of Economic Cooperation and Development (OECD). There were forty-three consolidated democracies in the world between 1979 and 1989.(12) Excluding the "mixed cases" of Switzerland and Finland, there were thirty-four parliamentary democracies, two semidential democracies, and only five pure presidential democracies.(13) We used the powerful yet relatively simple formula devised by Markku Laakso and Rein Taagepera to measure the "effective" number of political parties in the legislatures of these forty-one political systems.(14) Of the thirty-four parliamentary democracies, eleven had between three and seven effective political parties.(15) Both of the semipresidential democracies in this universe had between three and four effective political parties. However, no pure presidential democracy had more than 2.6 effective political parties. These data indicate that consolidated parliamentary and semipresidential democracies can be associated with a large number of parties in their legislatures, whereas consolidated presidential democracies are not associated with the type of multiparty coalitional behavior that facilitates democratic rule in contexts of numerous socio-economic, ideological, and ethnic cleavages and of numerous parties in the legislature. The currently empty column in Table 1 of long-standing presidential democracies with "3.0 or more" effective legislative parties is probably one of the reasons why there are so few continuous presidential democracies.

The Finnish political scientist Tatu Vanhanen published an important study of democratic durability that incorporates the nuances in individual countries, socioeconomic structures. Hence, it provides another data set for testing our hypothesis regarding constitutional frameworks.(16)

Vanhanen constructed a political Index of Democratization (ID) based on (1) the total percentage of the vote received by all parties except the largest vote getter and (2) the total percentage of the population that votes. He has also constructed a socioeconomic Index of Power Resources (IPR) based on six variables: (1) degree of decentralization of nonagricultural economic resources, (2) percentage of total agricultural land owned as family farms, and percentage of population (3) in universities, (4) in cities, (5) that is literate, and (6) that is not employed in agriculture. His major hypothesis is that all countries above his threshold level of 6.5 on his Index of Power Resources "should be democracies," and all countries below his minimum level, 3.5 index points, "should be non-democracies or semi-democracies." He has constructed his indexes for 147 countries for 1980 and 1988.

His hypothesis was broadly confirmed in that 73.6 percent of the countries that were above 6.5 in his IPR qualified as democracies as measured by his Index of Democracy. In his regression analysis with these indexes, Vanhanen found the correlation (r2) between the ID and IPR equal to .707 in 1980 and .709 in 1988. Approximately 76 percent of the 147 country cases tested by Vanhanen had small residuals and deviated from the regression line by less than one standard error of estimate.

However, thirty-six countries in 1980 and thirty-four in 1988 had negative or positive residuals larger than one standard error of estimate. These seventy large-residual cases indicate that about 24 percent of the variance in Vanhanen's regression analysis is unexplained. Vanhanen noted that "large positive residuals indicate that the level of democratization is considerably higher than expected on the basis of the average relationship between ID and IPR [we will call these cases `democratic over-achievers'], and large negative residuals indicate that it is lower than expected [we will call these `democratic underachievers']." He then asks "how to explain these deviations that contradict my hypothesis? I have not found any general explanation for them."(17)

Vanhanen's unexplained variance - his democratic over- and under-achievers - constitutes a data set with which to test our hypothesis regarding constitutional frameworks. Of the total seventy deviating cases in his 1980 and 1988 studies, fifty-nine occurred in constitutional frameworks we have called "pure parliamentary" or "pure presidential" (thirty-seven and twenty-two cases, respectively). When we analyze democratic underachievers in Vanhanen's set, we find that presidential systems had a democratic underachiever rate 3.4 times greater than did the parliamentary systems. Further, parliamentary systems in Vanhanen's set were 1.8 times more likely than presidential systems to be democratic overachievers. (See Table 2.)

Another set of data concerns both comparative capacity to be democratic survivors and vulnerability to military coups. Since we are concerned primarily with countries that are making some effort to construct democracies, we restrict our analysis to those countries in the world that qualified in the Gastil Political Rights Scale as democracies for at least one year between 1973 and 1989. Only 77 of the 168 countries in the world met this test. In an attempt to control for economic development as an intervening variable that might independently influence political stability, we eliminate from this section of our analysis the twenty-four OECD countries. This leaves a data set of the fifty-three NON-OECD countries that experimented with democracy for at least one year between 1973 and 1989. Of these, twenty-eight countries were pure parliamentary, twenty-five were pure presidential, and surprisingly none were either semipresidential or mixed. Only five of the twenty-five presidential democracies (20 percent) were democratic for any ten consecutive years in the 1973-89 period; but seventeen of the twenty-eight pure parliamentary regimes (61 percent) were democratic for a consecutive ten-year span in the same period. Parliamentary democracies had a rate of survival more than three times higher than that of presidential democracies. Pure presidential democracies were also more than twice as likely as pure parliamentary democracies to experience a military coup. (See Tables 3 and 4.)

[TABULAR DATA OMITTED]

Another source of relevant data concerns the set of countries, ninety-three in all, that became independent between 1945 and 1979.(18) During the ten-year period between 1945 and 1979 only fifteen of the ninety-three merit possible classification as continuous democracies. Since we are interested on evolution toward and consolidation of democracy, we examine the regime form that these countries chose at Independence. Forty-one countries functioned as parliamentary systems in their first year of Independence, thirty-six functioned as presidential systems, three functioned as semipresidential systems, and thirteen functioned as ruling monarchies. At this stage of our research, we are Impressed by the fact that no matter what their Initial constitutional form, not one of the fifty-two countries in the nonparliamentary categories evolved Into a continuous democracy for the 1980-89 sample period, whereas fifteen of the forty-one systems (46 percent) that actually functioned as parliamentary systems in their first year of independence not only evolved into continuous democracies but were the only countries in the entire set to do so. (See Table 5).

[TABULAR DATA OMITTED]

If the data in Table 5 were strictly numerical observations, the chances of this distribution occurring randomly would be less than one in one thousand. But we realize that the quantification of this qualitative data masks important realities, such as the fact that the classes catch some countries that were always a democratic or even antidemocratic. We do not rule out the hypothesis that the more democratic countries chose parliamentary systems at independence. Also, the fact that many of the "democratic survivors" are Island states and that all but two (Papua New Guinea and Nauru) are former British colonies should be taken into account.(19) We can control for the British colonial legacy, however, by Isolating the fifty former British colonies from our original set of ninety-three. Of the thirty-four from this subset that began independence as parliamentary systems, thirteen (38 percent) evolved into continuous democracies for the 1980-89 period. Of the five former British colonies that began as presidential systems, not one evolved into a democracy for the 1980-89 period.(20) Similarly, not one of the eleven former British colonies that began independence as ruling monarchies evolved Into a continuous democracy for INKS. This suggests that factors other than British colonial heritage are related to the democratic evolution and durability in these countries. Moreover, the fifteen democratic survivors in our set survived despite challenges such as tribal riots, linguistic conflicts, economic depressions, and/or mutinies. They therefore constitute a set of countries for which the constitutional form may be crucial in explaining democratic durability.

The comparative tendency for different constitutional frameworks to produce legislative majorities can also be ascertained. This is relevant to our central question because majorities help to implement policy programs democratically. Examining evidence from our set of the non-OECD countries that were democratic for at least one year from 1973 to 1987, we note that in presidential democracies the executive's party enjoyed a legislative majority less than half of the time (48 percent of the democratic years). Parliamentary democracies, in sharp contrast, had majorities at least 83 percent of the time. (See Table 6)

[TABULAR DATA OMITTED]

A final set of data concerns the duration and reappointment of cabinet ministers in presidential versus parliamentary frameworks. These data relate to the Issue of continuity in governance. Some minimal degree of ministers continuity and/or prior ministerial experience would seem to be helpful in enhancing the political capacity of the government of the day to negotiate with state bureaucracies and with national and transnational corporations. Using a number of recent studies, we have examined all ministerial appointments during the years of democratic rule in Western Europe, the United States, and Latin America between 1950 and 1980. Two major findings emerge. First, the "return ratio" of ministers (that is, the percentage who serve more than once on their careers) is almost three times higher in parliamentary democracies than in presidential democracies. Second, the average duration of a minister in any one appointment is almost twice as long in parliamentary democracies as it is in presidential democracies. Even when only those countries with more than twenty-five years of uninterrupted democracy are included in the sample, the findings still hold.(21) The conclusion is inescapable: ministers in presidential democracies have far less experience than their counterparts in parliamentary democracies.

The Contrasting Logics of Pure Parliamentarianism

and Pure Presidentialism

Let us step back from the data for a brief note about the type of statements that can be made about political institutions and democratic consolidation. The status of statements about the impact of institutions is not causally determinative (A causes B) but probabilistic (A tends to be associated with B). For example, Maurice Duverger's well-known observation about electoral systems is a probabilistic proposition: it holds that systems with single-member districts and where a simple plurality wins the seat tend to produce two-party systems, whereas electoral systems with multimember districts and proportional representation tend to produce multiparty systems.(22) The fact that Austria and Canada are exceptions to his proposition is less important than the fact that nineteen of the twenty-one cases of uninterrupted democracy in postwar Industrialized countries conform to his proposition.(23)

A probabilistic proposition in politics is more than a statistical assertion. It entails the identification and explanation of the specific political processes that tend to produce the probabilistic results. And to establish even greater confidence in the proposition, one should examine case studies to explain whether and how the important hypothesized institutional characteristics actually came into play in individual cases.(24)

Whatever the constitutional framework, consolidating democracy outside of the industrialized core of the world is difficult and perilous. The quantitative evidence we have brought to bear on presidentialism and parliamentarianism would assume greater theoretical and political significance if a strong case could be made that the empirically evident propensities we have documented are the logical, indeed the predictable, result of the constitutional frameworks themselves. We believe that such a case can be made.

The essence of pure parliamentarianism is mutual dependence. From this defining condition a series of Incentives and decision rules for creating and maintaining single-party or coalitional majorities, minimizing legislative impasses, inhibiting the executive from flouting the constitution, and discouraging political society's support for military coups predictability flows. The essence of pure presidentialism is mutual independence. From this defining (and confining) condition a series of incentives and decision rules for encouraging the emergence of minority governments, discouraging the formation of durable coalitions, maximizing legislative impasses, motivating executives to flout the constitution, and stimulating political society to call periodically for military coups predictability flows. Presidents and legislatures are directly elected and have their own fixed mandates. This mutual independence creates the possibility of a political Impasse between the chief executive and the legislative body for which there is no constitutionally available Impasse-breaking device.

Here, then, is a paradox. Many new democracies select presidentialism because they believe it to be a strong form of executive government. Yet our data show that between 1973 and 1987 presidential democracies enjoyed legislative majorities less than half of the time. with this relatively low percentage of supported time,, and the fixed mandates of the presidential framework, executives and legislatures in these countries were "stuck" with one another, and executives were condemned to serve out their terms. How often did these executives find it necessary to govern by decree-law - at the edge of constitutionalism - in order to Implement the economic restructuring and austerity plans they considered necessary for their development projects?

Our evidence shows that, in contrast to presidentialism, the executive's party In parliamentary democracies enjoyed a majority of seats in the legislature over 83 percent of the time period under study. For the remaining 17 percent of the years, parliamentary executives, motivated by the necessity to survive votes of confidence, formed coalition governments and party alliances in order to attract necessary support. When they were unable to do this, the absence of fixed mandates and the safety devices of the parliamentary Institutional framework allowed for calling rapid new elections, the constitutional removal of unpopular, unsupported governments through the vote of no confidence, or simply the withdrawal from the government of a vital coalition partner.

Parliamentarianism entails mutual dependence. The prime minister and his or her government cannot survive without at least the passive support of a legislative majority. The inherent mechanisms of parliamentarianism Involved in the mutual dependency relationship - the executives right to dissolve parliament and the legislature's right to pass a vote of no confidence - are deadlock-breaking devices. These decision rules do not assure that any particular government will be efficient in formulating policies nor do they assure government stability. But the decision mechanisms available in the parliamentary framework do provide constitutional means for removing deadlocked or inefficient governments (executives and parliaments. The danger that a government without a majority will rule by decree is sharply curtailed by the decision rule that allows the parliamentary majority (or the prime minister's coalition allies or even his or her own party) to call for government reformation.

Why is it logical and predictable that military coups are much more likely in pure presidential constitutional frameworks than in pure parliamentary frameworks? Because, as we discussed above, parliamentary democracies have two decision rules that help resolve crises of the government before they become crises of the regime. First, a government cannot form unless it has acquired at least a supported minority,, in the legislature second, a government that is perceived to have lost the confidence of the legislature can be voted out of office by the simple political vote of no confidence for in Germany and Spain by a positive legislative vote for an alternative government). Presidentialism, in sharp contrast, systematically contributes to impasses and democratic breakdown. Because the president and the legislature have separate and fixed mandates, and because presidents more than half of the time find themselves frustrated in the exercise of their power due to their lack of a legislative majority, presidents may often be tempted to bypass the legislature and rule by decree-law. It is extremely difficult to remove even a president who has virtually no consensual support in the country or who is acting unconstitutionally; it usually requires a political-legal-criminal trial (impeachment), whose successful execution requires exceptional majorities.(25) Thus, even when the socioeconomic crises are Identical in two countries, the country with the presidential system is more likely to find itself in a crisis of governance and will find it more difficult to solve the crisis before it becomes a regime crisis.(26) Such situations often cause both the president and the opposition to seek military involvement to resolve the crisis in their favor.

Guillermo O'Donnell documented a phenomenon observed in the new Latin American democracies in his extremely interesting (and alarming) article on "delegative democracy," a conceptual opposite of representative democracy.(27) Key characteristics of delegative democracy include (1) presidents who present themselves as being "above" parties, (2) institutions such as congress and the judiciary that are viewed as "a nuisance," with accountability to them considered an unnecessary impediment, (3) a president and his staff who are the alpha and omega of politics, and (4) a president who Insulates himself from most existing political institutions and organized interactions and becomes the sole person responsible for "his" policies. We suggest that these characteristics of O'Donnell's delegative democracy are some of the predictable pathologies produced by the multiple logics of the presidential framework. Consider the following: Presidential democracy, due to the logic of its framework, always produces (1) presidents who are directly elected and (2) presidents with fixed terms. Presidential democracy often produces (1) presidents who feel they have a personal mandate and (2) presidents who do not have legislative majorities. Thus, the logic of presidentialism has a strong tendency to produce (1) presidents who adopt a discourse that attacks a key part of political society (the legislature and parties) and (2) presidents who increasingly attempt to rely upon a "state-people" political style and discourse that marginalizes organized groups in political society and civil society. Delegative democracy can no doubt exist in the other constitutional frameworks; however, the multiple logics of pure parliamentarianism seem to work against delegative democracy.

Why are there many enduring multiparty parliamentary democracies but no long-standing presidential ones? In a parliamentary system, the junior political parties that participate in the ruling coalition are Institutional members of the government and are often able to negotiate not the ministries they will receive, but who will be appointed to them. All members of the coalition have an incentive to cooperate if they do not want the government of the day to fall. In these circumstances, democracies with four, five, or six political in the legislature can function quite well.

There are far fewer incentives for coalitional cooperation in presidentialism. The office of the presidency is nondivisible. The president may select members of the political parties other than his own to serve in the cabinet, but they are selected as individuals, not as members of an enduring and disciplined coalition. Thus, if the president's party (as in President Collor's party in Brazil) has less than 10 percent of the seats in the legislature, he rules with a permanent minority and with weak coalitional incentives. On a vote-by-vote basis, the president may cajole or buy a majority, but repeated purchases of majorities are absolutely inconsistent with the principled austerity plans of restructuring that face most East European and Latin American democracies.

East European or Latin American political leaders who believe that their countries, for historical reasons, are inevitably multiparty in political representation are playing against great odds if they select a presidential system, as the existing evidence demonstrates. Brazil's high party fragmentation, for example, has contributed to a presidential-legislative deadlock that has frozen the lawmaking process in an already fragile democracy. Party fragmentation, the lack of party discipline, and general party underdevelopment in Brazil have been exacerbated by its electoral system, which combines proportional representation with an open list. The 1990 elections yielded 8.5 effective parties in the Brazilian Chamber of Deputies and 6.0 in the Senate.(28) These numbers seem alarmingly high considering that all the long-standing, pure presidential democracies reported in Table 1 had fewer than 2.6 effective political parties.

Moreover, the closer a country approaches the ideal types of "sultanship," "totalitarianism," or early "posttotalitarianism" the "flatter" are their civil and political societies.(29) In these circumstances, adopting the constitutional framework of presidentialism in the period of transition from sultanship, totalitarianism, or early posttotalitarianism reduces the degrees of freedom for an emerging civil and political society to make a midcourse correction, because heads of government have been elected for fixed terms (as in Georgia). In contrast, the Bulgarian transition had significant parliamentary features, which allowed an emerging political society to change the prime minister (and the Indirectly elected president) so as to accommodate new demands.

In Poland, where constitutional reformers are flirting with the idea of strengthening the role of the president, party fragmentation is even greater than in Brazil, the effective number of parties in the Polish Sejm after the 1991 legislative elections was 10.8.(30) Most of these parties in the Polish legislature, like those in Brazil, lack clear programs and exist as mere labels for politicians to use for election into office.(31) Our data suggest that Poland would be playing against the odds were it to move to toward a purely presidential system.

Also flowing from the logic of the constitutional framework are the questions of why ministers serve short terms in presidential democracies and why they are rarely reappointed in their lifetime. Because presidents do not normally enjoy majorities in the legislature, they resort to rapid ministerial rotation as a device in their perpetual search for support on key issues. In parliamentary systems, by contrast, coalitional majorities make such rapid turnover unnecessary. Furthermore, key ministers usually have long and strong associations with their political parties and are often reappointed as government coalitions form and re-form during the life of their careers. In presidential democracies, ministers are strongly associated with a particular president, leave office when the president does, and normally never serve as a minister again in their life.

CONCLUSION

Let us consider the question that follows from the data. Why does pure parliamentarianism seem to present a more supportive evolutionary framework for consolidating democracy than pure presidentialism? We believe we are now in a position to say that the explanation of why parliamentarianism is a more supportive constitutional framework lies in the following theoretically predictable and empirically observable tendencies: its greater propensity for governments to have majorities to implement their programs; its greater ability to rule in a multiparty setting; Its lower propensity for executives to rule at the edge of the constitution and its greater facility at removing a chief executive who does so; its lower susceptibility to military coup; and its greater tendency to provide long party-government careers, which add loyalty and experience to political society.

The analytically separable propensities of parliamentarianism interact to form a mutually supporting system. This system, qua system, increases the degrees of freedom politicians have as they attempt to consolidate democracy. The analytically separable propensities of presidentialism also form a highly Interactive system, but they work to impede democratic consolidation.

(*) This article grew out of an exchange at a December 1990 meeting in Budapest of the East-South System Transformations Project, which brought together specialists on Eastern Europe, Southern Europe, and South America. When we were discussing topics for future research and dividing up our collective work, Adam Przeworski lamented that although there were assertions in the literature about the probable impact of different types of institutional arrangements on democratic consolidation, there were no systematic data available. In his notes about the Budapest meeting, Przeworski reiterated that "we seem to know surprisingly little about the effects of the particular institutional arrangements for their effectiveness and their durability. Indeed, the very question whether institutions matter is wide open." See Przeworski, "Notes after the Budapest Meeting" (Chicago: University of Chicago, January 11, 1991), 10. We acknowledge the careful reading and/or comments of Adam Przeworski, Jack Snyder, Douglas Rae, Juan Linz, Mike Alvarez, Martin Gargiulo, Lisa Anderson, Anthony Marx, Gregory Gause, Joel Hellman, and Scott Mainwaring. The normal caveats apply.

(1) See, e.g., Stephan Haggard and Robert R. Kaufman, eds., The Politics of Economic Adjustment (Princeton: Princeton University Press, 1992) Adam Przeworski, Democracy and the Market: Political and Economic Reforms in Eastern Europe and Latin America (Cambridge: Cambridge University Press, 1991) and Christopher Clague and Gordon C. Rausser, eds., The Emergence of Market Economies in Eastern Europe (Cambridge, England: Blackwell Press, 1992). (2) James G. March and Johan P. Olsen, "The New Institutionalism: Organizational Factors in Political Life," American Political Science Review 78 (September 1984), 738. For a pioneering early work exemplifying this approach, see Maurice Duverger, Political Parties (New York: Wiley, 1954). Other important works that explore the causal relationship between institutions such as electoral systems and political parties, and democratic stability include Giovanni Sartori, Parties and Party Systems: A Framework for Analysis (Cambridge: Cambridge University Press, 1976); Douglas Rae, The Political Consequences of Electoral Laws (New Haven: Yale University Press, 1967); William H. Riker, The Theory of Political Coalitions (New Haven: Yale University Press, 1962); Bernard Grofman and Arend Lijphart, eds., Electoral Laws and Their Political Consequences (New York: Agathon, 1986); Rein Taagepera and Matthew Soberg Shugart, Seats and Votes knew Haven: Yale University Press, 1989); and Matthew Soberg Shugart and John Carey, Presidents and Assemblies (Cambridge: Cambridge University Press, 1992). An important work in the neo-institutionalist literature that focuses on legislatures and structure-induced equilibrium is Kenneth Shepsle, "Institutional Equilibrium and Equilibrium Institutions," in Herbert F. Weisberg, ed., Political Science: The Science of Politics knew York: Agathon, 1986). See also Mathew D. McCubbins and Terry Sullivan, eds., Congress: Structure and Policy (Cambridge: Cambridge University Press, 1987).
(3) There is a growing literature on this question. Much of it is brought together in Juan J. Linz and Arturo Valenzuela, eds., Presidentialism and Parliamentarianism: Does It Make a Difference? (Baltimore: Johns Hopkins University Press, forthcoming). However, no article in this valuable collection attempts to gather systematic global quantitative data to address directly the question raised in the title of the book and by Przeworski. Linz first appeared in print on this subject in a brief "Excursus on Presidential and Parliamentary Democracy," in Linz and Alfred Stepan, eds., The Breakdown of Democratic Regimes (Baltimore: Johns Hopkins University Press, 1978). His much-cited seminal "underground" paper with the same title as his forthcoming book was first presented at the workshop on "Political Parties in the Southern Cone," Woodrow Wilson International Center, Washington, D.C., 1984; see also idem, "The Perils of Presidentialism," Journal of Democracy 1 (Winter 1990). See also Scott Mainwaring, "Presidentialism, Multiparty Systems, and Democracy: The Difficult Equation," Kellogg Institute Working Paper, no. 144 (Notre Dame, Ind.: University of Notre Dame, September 1990).
(4) We agree with Philippe C. Schmitter's argument that there are many types of democracies and that "consolidation includes a mix of institutions." See Schmitter, "The Consolidation of Democracy and the Choice of Institutions," East-South System Transformations Working Paper, no. 7 (Chicago: Department of Political Science, University of Chicago, September 1991), 7. See also Schmitter and Terry Karl, "What Democracy Is ... and Is Not," Journal of Democracy (Summer 1991). The authors list eleven important dimensions that provide a matrix of potential combinations by which political systems can be differently democratic.
(5) We realize that any effort to operationalize the concept of "democracy" so that it can be used for purposes of classification of all the countries of the world is inherently difficult. Fortunately there have been two independently designed efforts that attempt this task. One, by Michael Coppedge and Wolfgang Reinicke, attempted to operationalize the eight "institutional guarantees" that Robert Dahl argued were required for a polyarchy. The authors assigned values to 137 countries on a polyarchy scale, based on their assessment of political conditions as of mid-1985. The results are available in Coppedge and Reinicke, "A Measure of Polyarchy" (Paper presented at the Conference on Measuring Democracy, Hoover Institution, Stanford University, May 27-28, 1988); and in idem, "A Scale of Polyarchy," in Raymond D. Gastil, ed., Freedom in the World: Political Rights and Civil Liberties, 1987-1988 (New York: Freedom House, 1990), 101-28. Robert A. Dahl's seminal discussion of the institutional guarantees needed for polyarchy is found in his Polyarchy: Participation and Opposition (New Haven: Yale University Press, 1971), 1-16. The other effort to operationalize a scale of democracy is the annual Freedom House evaluation of virtually all the countries of the world. The advisory panel in recent years has included such scholars as Seymour Martin Lipset, Giovanni Sartori, and Lucian W. Pye. The value assigned for each year 1973 to 1987 can be found in the above-cited Gastil, 54-65. In this essay, we will call a "country a "continuous democracy" if it has received no higher than a a scale score of 3 on the Coppedge-Reinicke Polyarchy Scale for 1985 and no higher than a 2.5 averaged score of the ratings for "political rights" and "civil liberties" on the Gastil Democracy Scale, for the 1980-89 period.
(6) On the defining characteristics of semipresidentialism, see the seminal article by Maurice Duverger, "A New Political System Model: Semi-presidential Government," European Journal of political Research 8 (June 1980). See also idem, Echec au Roi (Paris: Albin Michel, 1978); and idem, Le monarchie republicaine (Paris: R. Laffont, 1974).
(7) For a discussion of the semipresidential constitutional framework, its inherent problem of "executive dualism," and the exceptional circumstances that allowed France to manage these problems, see Alfred Stepan and Ezra N. Suleiman, "The French Fifth Republic: A Model for Import? Reflections on Poland and Brazil," in H. E. Chehabi and Alfred Stepan, eds., Politics, Society and Democracy: Comparative Studies (Boulder, Colo.: Westview Press, forthcoming).

(8) Alfred Stepan will develop this argument in greater detail in a book he is writing entitled Democratic Capacities/Democratic Institutions.
(9) For example, in Arend Lijphart's list of the twenty-one continuous democracies of the world since World War II, seventeen were pure parliamentary democracies, two were mixed, one was semipresidential, and only one, the United States, was pure presidential. See Lijphart, Democracies: Patterns of Majoritarian and Consensus Government in Twenty-one Countries (New Haven: Yale University Press, 1984), 38

(10) The norm is a directly elected president with very strong de jure and de facto prerogatives coexisting with a prime minister who needs the support of parliament. As of this writing (April 1993), only Hungary and the newly created Czech Republic and Slovakia had opted for the pure parliamentary constitutional framework. Despite having directly elected presidents, Slovenia, Estonia, and Bulgaria have strong parliamentary features. In Slovakia and Estonia presidents will now be selected by parliament. Bulgaria, however, has moved from an indirectly to a directly elected president. For political, legal, and sociological analyses of constitution making in East European transitions, see the quarterly publication East European Constitutional Review, which is part of the Center for the Study of Constitutionalism in Eastern Europe at the University of Chicago. The center was established in 1990 in partnership with the Central European University.

(11)Duration analysis would be particularly appropriate because it estimates the conditional probability of an event taking place (for example, of a democracy "dying," by undergoing military coup), given that the regime has survived for a given period of time as a democracy. This conditional probability is in turn parameterized as a function of exogenous explanatory variables (such as constitutional frameworks). The sign of an estimated coefficient then indicates the direction of the effect of the explanatory variable on the conditional probability of a democracy dying at a given time. Such models allow us to estimate whether democracies exhibit positive or negative "duration dependence": specifically, whether the probability of a democracy dying increases or decreases, respectively, with increases in the duration of the spell. Mike Alvarez, a Ph.d. candidate in political science at the University of Chicago, is creating the data and the appropriate statistical techniques and then implementing this duration analysis as part of his dissertation. Adam Przeworski, too, has embarked on such research. See also Nicholas M. Kiefer, "Economic Duration Data and Hazard Functions," Journal of Economic Literature 26(June 1988).

(12) We consider a country to be a "consolidated democracy" if it has received no higher than a scale score of 3 on the Coppedge-reinicke Polyarchy Scale for 1985 and no higher than a 2.5 average of the ratings for "political rights" and "civil liberties" on the Gastil Democracy Scale. Countries that met these joint criteria for every year of the 1979-89 decade are considered "continuous consolidated democracies." See fn. 18 herein.

(13) Duverger calls Finland semipresidential because the president has significant de jure and de facto powers; it should be pointed out, however, that from 1925 to 1988 the Finnish president was not so much directly elected as indirectly chosen by party blocs. The candidates normally did not campaign in the country, and though parties put the names of their candidates on the ballot, the electoral college votes were not pledges and often entailed deliberations and multiple balloting, leading Shugart and Carey to conclude that the presidential election system in Finland from 1925 to 1988, "given its party-centered character ... was not much different from election in parliament." See Shugart and Carey (fn 2), 212-21, 226-28, quote at 221. We consider Finland to have been a "mixed" constitutional system until 1988.

(14) Laakso and Taagepera, "Effective' Number of Parties: A Measure with Application to West Europe," Comparative Political Studies 12(April 1979). The formula takes into account each party's relative size in the legislature, as measured by the percentage of seats it holds. The "effective" number of parties is "the number of hypothetical equal-size parties that would have the same total effect on fractionalization of the system as have the actual parties of unequal size." The formula for calculating the effective number of parties (N) is

[MATHEMATICAL EXPRESSION OMITTED]

where pi = the percentage of total seats held in the legislature by the i-th party.

For each country listed in Table I, we determined the number of seats held in the lower or only house of the legislature at the time of each legislative election between 1979 and 1989 Then, the effective number of political parties (N) was calculated for each of these election years and multiplied by the number of years until the next legislative election. (15) Austria, Ireland, and Iceland have directly elected presidents, but we do not classify them as semipresidential; we concur with Duverger that they are not de facto semipresidential since political practice is parliamentary." See Duverger (fn. 6, 1980), 67. (16) See Vanhanen, The Process of Democratization: A Comparative Study of 147 States, 1980-1988 (New York: Crane Russak, 1990).

(17) lbid., 84.

(18) We use the date of independence since it was usually within one year of independence that new constitutions were drafted and approved in these countries. We exclude from our analysis those countries that became independent after 1979 because we want to see which of these countries were then continuously democratic for the ten-year period 1980-89. This gives us a sample of time between World War II and 1979.

(19) Myron Weiner observes that most of the smaller, newly independent democracies ... are also former British colonies,, and puts forth the hypothesis that "tutelary democracy under British colonialism appears to be a significant determinant of democracy in the Third World." See Weiner, "Empirical Democratic Theory," in Myron Weiner and Ergun Ozbudun, eds., Competitive Elections in Developing Countries (Durham, N.C.: Duke University Press, 1978), esp. 18-23, quote at 19. This question is also addressed by Jorge Dominguez, "The Caribbean Question: Why Has Liberal Democracy (Surprisingly) Flourished?" in ed., Democracy in the Caribbean: Political Economic, and Social Perspectives (Baltimore Hopkins University Press, 1993). Dominguez discusses how these Caribbean democracies have faced and survived I severe economic crises. He attributes their democratic stability to the legacy of British institutions (including, but not limited to, the Westminster parliamentary model) and the prodemocratic disposition of the countries, leadership.

(20) The five former British colonies that chose presidential systems within one year of independence were Zambia, Cyprus, Malawi, Seychelles, and South Yemen.

(21) See Jean Blondel, Government Ministers in the Contemporary World (Beverly Hills, Calif.: Sage, 1985), esp. appendix II 277-81, Mattei Dogan, Pathways to Power: Selecting Rulers in Pluralist Democracies (Boulder, Colo.: Westview Press, 1989). Waldino C. Suarez: "Argentina Political Transition and Institutional Weakness in Comparative Perspective," in Enrique A. Baloyra, ed., Comparing New, Democracies.. Transition and Consolidation in Mediteranian Europe and the Southern Cone (Boulder, Colo.: Westview Press, 1987 idem, "El gabinete en America Latina: Organizacion y cambio," Contribuciones, no. 1 (January-March 1985); and idem, "El Poder ejecutivo en America Latina: Su capacidad operativa bajo regimenes presidentialistas de gobierno," Revista de Estudios Politicos, no. 29 (September-October 1982).

(22) See Duverger (fn. 2)

(23) For a discussion of Duverger's proposition in the context of modern industrialized democracies see Arend Lijphart, Democracies: Patterns of Majoritarian and Consensus Government in Twenty-one Countries (New Haven: Yale University Press, 1984), 156-59.

(24) There is a growing literature of case studies examining the influence of constitutional frameworks on stability and/or breakdown in developing countries. See, e.g., David M. Lipset, "Papua New Guinea: "The Melanesian Ethic and the Spirit of Capitalism, in Larry Diamond, Juan J. Linz, and Seymour Martin Lipset, eds., Democracy in Developing Countries.. Asia (Boulder, Colo.: Lynne Rienner, 1989 esp. 413. Lipset discusses how the constitutional framework came into play to prevent regime breakdown in Papua New Guinea. See also Dominguez (fn. 19).

(25) Schmitter and Karl (fn. 4) quite correctly build into their definition of democracy the concept of accountability. But with the exception of the U.S. where a president can be directly reelected only once, no president in any other long-standing democracy in the world, once in office, can be held politically accountable by a vote of the citizens, representatives accountability mechanism is so extreme and difficult - with the political-legal-criminal trial that needs exceptional majorities (impeachment) that the accountability principle in presidentialism is weaker than in parliamentarianism.

(26) For theoretical differentiation between crises of government and crises of regime, see Juan J. Linz and Alfred Stepan, eds., The Breakdown of Democratic Regimes (Baltimore: Johns Hopkins University Press, 1978, esp. 74.

(27) See O'Donnell, "Democracia Delegativa?" Novos Estudos CEBRAP, no. 31 (October 1991). (28) These numbers were calculated using the Laakso/Taagepera formula and the data reported reported in Keesings Record of World Events (1990) and Arthur S. Banks, ed., Political Handbook book of the World (Binghamton); CSA Publishers, State University of New York at Binghamton 1991.

(29) This argument is developed in Juan J. Linz and Alfred Stepan, "Problems of Democratic Transition and Consolidation: Eastern Europe, Southern Europe and South America" (Book manuscript), pt. 1.

(30) This is developed in Stepan and Suleiman (fn. 7).

(31) For a discussion of how both the political culture and the institutional structure in Brazil contributed to the country's weak party system, see Scott Mainwaring, Dilemmas of Multiparty Presidential Democracy: The Case of Brazil," Kellogg Institute Working Paper no. 174 (Notre Dame, Ind.: University of Notre Dame, 1992). See also idem, "Politicians, Parties, and Electoral Systems: Brazil in Comparative Perspective," Comparative Politics 24 (October 1991); and his forthcoming book on Brazilian political parties.