THE WEIGHT OF HISTORY AND THE REBUILDING OF BRAZILIAN DEMOCRACY *

Nation-states, like their citizens, are dependent on their peers for cues about what passes for appropriate behavior. With respect to the evolution of political institutions, the result is the proliferation of political fads among certain clusters of countries, a process which scholars describe in terms of “waves”, “contagion”, “isomorphism”, and “refl ection”. The idea that the adoption of a practice by one actor would infl uence the probability of adoption by another (a useful defi nition of the concept diffusion1) is an intriguing metasubject of inquiry within a variety of disciplines and a variety


Zachary Elkins
Nation-states, like their citizens, are dependent on their peers for cues about what passes for appropriate behavior.With respect to the evolution of political institutions, the result is the proliferation of political fads among certain clusters of countries, a process which scholars describe in terms of "waves", "contagion", "isomorphism", and "refl ection".The idea that the adoption of a practice by one actor would infl uence the probability of adoption by another (a useful defi nition of the concept diffusion1 ) is an intriguing metasubject of inquiry within a variety of disciplines and a variety The weight of history and the rebuilding of Brazilian democracy of topics2 .In recent years, scholars of democracy -noting the symptoms of diffusion in their own phenomenon of interest -have produced an inspirational set of studies on the process (Starr, 1991; Markoff, 1996; O'Laughlin, Ward, et al., 1998;  Coppedge and Brinks, 1999).These studies provide much needed confi rmation, and in some cases, elaboration, of a powerful mechanism of institutional change.
At this point in the research cycle, we can be confi dent that institutional and policy transitions are highly contagious.The intent of the analysis below is to document the process of contagion at the level of the decision maker in transitioning countries.While the study is grounded in, and inspired by, evidence of contagion in democratization, we turn our attention away from the choice of regime typea choice that is usually not debated publicly or concretely.Rather, the analysis below focuses on a very particular, but fundamental, institutional choice that confronts actors in new democracies: whether to adopt a presidential or parliamentary system of government.Anecdotal evidence and intuition suggests that this decision is highly dependent on the decisions of neighboring and otherwise relevant governments.I examine the decision process in Brazil, an important and recent case of democratization.This country is especially interesting because of its historical experience with both parliamentarism and presidentialism, the comprehensive agenda of its recent constitutional convention, and its strong ties to both the Americas and Europe -two important regions with opposing systems Zachary Elkins of government.The analysis below reviews evidence at both the cross-national level and the individual level that suggests that 1. foreign experience and foreign models are remarkably infl uential and relevant to the designers of policies in new democracies; 2. high-achieving countries as well as culturally similar countries make for infl uential models; and 3. institutional choice is highly path dependent and resistant to innovation.

The distribution of parliamentarism and presidentialism worldwide
Systems of government are highly segregated by culture, geography, and economic achievement.As the 1997 map in Figure 1 demonstrates vividly, regions of the world tend to be either parliamentarist or presidentialist but not both.The only regions with much diversity are Asia and Eastern Europe and even these are an overwhelming 70% presidentialist and parliamentarist, respectively.The weight of history and the rebuilding of Brazilian democracy Of course, geography is strongly related to cultural characteristics like language, religion, and colonial heritage.In fact, Table 1 suggests that diversity within regions can be attributed to differences in colonial heritage, language, and religion.For example, while the Americas are largely presidentialist, the former British colonies in the region are universally parliamentarist with the exception of the United States.Similarly, Eastern Europe is predominately parliamentarist except for the states of the former Soviet Union, which are, to a country, presidentialist.Along with language and colonial heritage, religion makes for an excellent marker of cultural identity as well, and displays the same degree of institutional homogeneity.Observe, for example, the almost universal presidentialism among largely Muslim nations.However, all this is not to say that religion and blood are always thicker than geography.The former British possessions in Africa look like their presidential neighbors and not their former protector and patron.What seems clear, without turning to more formal multivariate analysis, is that systems of government are highly dependent on geography and several markers of culture.
Parliamentarist and presidential governments differ markedly also by their economic resources and achievements.As Table 1 shows, parliamentary systems are, on average, signifi cantly more developed.Countries with parliamentary systems have a higher Gross Domestic Product (GDP) per capita, a higher degree of economic equality, a higher life expectancy, and a higher credit rating than do those with a presidential system.This discrepancy is even more pronounced if we exclude the United States.
These profi les are not intended as historical evidence on the diffusion of different systems across the globe.Our interest at this point, however, is simply to document the currently highly clustered nature of systems of government in order to describe the unique setting in The weight of history and the rebuilding of Brazilian democracy which governments now operate.Nevertheless, it seems safe to assert that the distribution of systems results from a pattern of both imitation among peers and correlated, but independent, decisions by governments with similar histories and structural assets3 .

Five propositions about continuity and change in systems of government
What are the forces for, and against, institutional change in an environment in which the distribution of choices is highly clustered along cultural, economic, and geographic lines?Our intuition is that such an environment, in which there are clear policy signals from very cohesive reference groups, provides strong incentives to conform to group norms.The expectation is that these external infl uences are at least as strong as any domestic impulses and calculations for change.We begin with fi ve propositions.
Proposition 1: Governments will be reluctant to deviate from the practices of their cultural and geographic peers.
Why should neighbors and cultural peers be so infl uential?One reason is that similar or adjacent entities will interact more often.More contact and communication results in more shared information about practices.Axelrod  (1997, p. 205) develops a model of the dissemination of culture that abstracts from this fundamental principle to say that communication is most effective between "similar" people.His theory of the diffusion of ideas specifi es mechanisms of change for local actors in the absence of any coordinating central authority.In his model, actors Zachary Elkins are adaptive rather than fully rational: they follow simple rules about giving and receiving infl uence, but they do not necessarily calculate costs and benefi ts in a strategic, forward-looking way.The result of Axelrod's model are pockets of ideational convergence, based on the number of features that two neighbors share in common.In the case of nation-states, increased information about foreign practices translates into imitation in a number of ways.For example, foreign models can encourage or expedite adoption by inserting a policy on a legislature's agenda, by offering a ready-made answer to domestic pressure for "change" and "innovation", by legitimating conclusions or predispositions already held, or by adding a decisive data point in the evaluation of alternatives (Bennett  1991a, 1991b).
Increased communication among countries, however, is not the only motor behind social infl uence.Culturally similar entities, whether or not they communicate extensively, constitute a relevant reference group with an established code of behavior.As John Meyer and co-authors argue persuasively, nations and organizations are remarkably responsive to the need to conform to these norms (Meyer and Rowan, 1977).Rosenau (1990)  terms these reference groups "cathectic", suggesting that decision makers have a strong cultural sense of whom their nation should look like.In this sense, collectives may adopt institutions for symbolic or ceremonial reasons quite independent of effi ciency criteria (Powell  and DiMaggio, 1991).In the case of presidentialism and parliamentarism, in which the institutions are highly clustered along cultural and geographic lines, the expectation is that the mechanisms of communication and conformity will be especially strong.
Proposition 2: Governments will be attracted to the policies of more economically successful governments.
The weight of history and the rebuilding of Brazilian democracy One of the basic tenets of social infl uence is that actors of lower social status emulate those of higher status.Policy makers might emulate the policies of successful, high achieving nations under the assumption that these nations possess some degree of expertise or even that their enhanced performance is in part due to their superior institutions.Westney (1987) makes this argument in her description of Japan's adoption of police, postal, and newspaper institutions from the West in the 1800's.In diffusion research, such transfers fall under the concept of hierarchical diffusion (see Lerner, 1964; Collier and Messick, 1975; Rogers,  1995).Since a strict class distinction between presidentialists and parliamentarists makes it very clear what the status structure will be, we expect the forces of hierarchical diffusion to be strong.
Proposition 3: Among developing nations, there will be some resistance towards the policies of a hegemonic or imperial power.
The proposition adds an important qualifier to proposition 2. Anti-imperialism is a strong feeling in developing nations and political policies have very symbolic power.While successful nations can serve as natural showcases with alluring models, success can breed as much resentment as it can admiration.
Proposition 4: Young states are more susceptible to external infl uences, and thus policy transition, than are older states.
Proposition 5: A government's institutional choice is dependent largely upon the generation of its birth.
These two propositions stem from the premise that institutional choices are extremely path dependent and, once adopted, hard to amend.A generation or two in a continued policy state can build in citizens a strong symbolic, and in the case of leaders, professional Zachary Elkins attachment, to their institution.This is a common observation in the literature on the diffusion of innovations and bears examination in the case of political institutions.The crucial insight here is that a symbolic attachment to institutions can thwart a transition to what experts may agree are superior, or at least more appropriate, institutions.The QWERTY typewriter and non-Metric systems are two examples of inferior practices that continue largely due to real or perceived costs of transition.This insight leads us to two interesting expectations for political institutions.First, it is fair to assume that in young states the accumulated attachment to institutions is low, and so the costs of removing existing structures and practices is similarly low.It is, therefore, these young, embryonic states that should be most sensitive to the infl uence of their cultural, geographic, and economic peers.Second, and consequently, we suggest that the evolution of a government's system of government is highly dependent on the prevailing wisdom during the era of its birth.As such, we should see distinct "generational" differences among governments with respect to parliamentarism and presidentialism, according the government's date of birth.In this paper, we merely note these propositions; our evidence focuses on the fi rst three.

Baseline Domestic Political Predictors of Institutional Transition
While we are most interested in external infl uences on a government's choice of institution, we must also consider internal forces unrelated to a government's international and generational environment.The assumption behind such explanations is that decision makers are largely unaffected by the infl uence of their peers.When crises arise that precipitate a reconsideration of their policies and institutions, they either look to the experience of their own The weight of history and the rebuilding of Brazilian democracy nation or act purely from their own preferences and needs.Their own preferences and needs have to do mainly with the structure of their other political institutions.
There are at least three plausible structural reasons for leaders to support one system over the other.First, parliamentarism thrives in nation states that have strong and unifi ed national parties.Which characteristic leads to the other is unclear, although it is probable that the installation of parliamentarism leads to these strong parties.It is also plausible that governments characterized by weak parties will view presidentialism as a better fi t than they will parliamentarism.Second is the related fi nding that very few large federal states have parliamentarism.While parliamentarism is certainly possible in a federal state (e.g., Germany), it is seems reasonable to think that diversity and decentralization in such states makes it diffi cult to build the strong and unifi ed national parties that parliamentarism requires.Third, the decision process of some leaders will undoubtedly include an estimate of the probability that either system will advance their professional interests.Presumably, they would be in favor of whichever system would be most amenable to electing, as national executive, themselves or their preferred candidate (or, conversely, avoiding the election of their least preferred candidate).These domestic factors serve as benchmark explanations by which we judge the strength of the external infl uences.

Rates of transition between systems
The literature proliferating in the last decade on the merits of presidentialism and parliamentarism has a prescriptive fl avor that implies reasonable odds of transition between systems (Linz, 1990, 1994; Stepan  and Skach, 1993; Riggs, 1997).The reality is that such transitions are rare.In fact, the records of one leading Zachary Elkins dataset (Alvarez et al., 1999) do not contain a single case of transition between the two systems between 1950 and 1990 (the time period covered by the data)4 .To some degree, such continuity is a methodological artifact of the Alvarez et al. (1999), which do not differentiate between parliamentary and presidential systems in cases that they deem non-democratic (a category which includes 60% of the cases in the sample).
Notwithstanding the Alvarez et al. (1999) coding decision, system-of-government transitions are a rare event.A useful World Bank dataset, which covers the period 1975 to 1987 and excludes far fewer nondemocracies (14.7% are coded non-democracies), shows roughly one or two transitions a year in each direction.Table 2 reports these transition probabilities for shifts in each direction and identifi es the cases of transition.Note that these rates should be considered an upper limit since some transitions, like that of Spain and Portugal to parliamentarism, might be better understood as transitions from authoritarianism than from presidential democracy.Nevertheless, transitions in either direction (the rates of which are less than 2%), are a rare event by most standards.For comparison, they are similar to those for transitions to democracy, which in the last fi fty years average a little less than 2%, and about half the rates of transitions to liberal economic policies which tend to occur about 4 to 5 % of the time (Simmons and Elkins, 2003) 5 .
The weight of history and the rebuilding of Brazilian democracy

Parliamentarism and presidentialism in Brazil
Given the rarity of these transitions, there is much to be gained by looking carefully at the decision process in a particular case where such a transition is under consideration.Brazil makes for an intriguing case study for several reasons.The country is unique in the Americas in that it has extensive experience with both presidentialism and parliamentarism.Immediately following independence from Portugal, Brazil existed with a semi-parliamentary system until the deposition of emperor Dom Pedro II in 1889 6 .At that point, Brazil gravitated to a presidential republic fashioned after the United States version 7 .Throughout the century, 6 Brazilian independence, itself, was a unique affair.Dom Pedro I, the son of the Portuguese king, residing in Rio de Janeiro, himself declared Brazil's independence.This irony was the result of the king's having left Lisbon for Rio in order to seek refuge from Napoleon, and then -comfortably ensconced in Rio and fi nding it more and more diffi cult to rule Portugal remotely -cutting ties with Lisbon.
With respect to the system of government, this had unique consequences for Brazil.In order to accommodate a royal head of state, but still adopt a more representative system -as the powerful ideas from the American and French revolutions demanded -Brazil adopted a semi-parliamentary system. 7A stable parliamentary system continued in Brazil until Dom Pedro II's deposition in 1889.With the end of the empire, leaders convened a constitutional assembly to devise a new set of rules.From the results of the 1891 Constitutional Assembly, it is clear that the young United States served as the principal model for Rui Barbosa and other founders of the Brazilian republic.The founders commissioned three authors who, working independently, were to draft initial versions from which to craft the fi nal document.Not only did all three produce a presidential plan, but also all three employed language from the US constitution to do so.Compare Americo Brasiliense's version "The exercise of executive power of the federation will be conferred on a single person who will have the title of President of the United States of Brazil; his mandate will be for four years" (Franco and Pilla, 1958)  with the United States version "The executive power shall be vested in a President of the United States of America.He shall hold his offi ce for the term of four years" (Article II, Section I of the US Constitution).Even the new name of the federation, United States of Brazil, was no accidental reference to the northern US.The adoption process was as clear as could be.Upon receiving the "new" constitution, one delegate wrote to a friend, "we all knew that it was not an original work or any sort political experimentation.[The three drafters] presented us with the text of the North American constitution, completed with a few lines from the Swiss and Argentine documents" (Amaro Cavalcanti apud Franco and Pilla, 1958).
The turn away from Europe, or more exactly, towards the United States had been a number of years in the making.Clearly, the young United States model had The weight of history and the rebuilding of Brazilian democracy however, parliamentarism has had adherents who have presented periodic proposals for its reinstatement 8 .Brazilian leaders even reverted to parliamentarism for one brief stint, in 1961, during a time of crisis 9 .
Brazil is an interesting case also because of its close ties to both the United States and Europe -the world's two principal models of presidentialism and parliamentarism respectively.Economically, Brazil depends upon the United States for most of its trade and external capital.Politically and culturally, however, Brazilians are very cognizant of their European roots and often prefer European products much to recommend it.Intellectually, it represented a fresh, progressive answer for other countries in the hemisphere that did not yet have reason to resent North American power and infl uence.To many it was still an irreverent answer to centuries of European domination on the continent.More importantly, it was the model adopted by each and every one of the newly independent Latin American states.It was clear that Brazil was conscious of not fi tting in with its neighbors.As early as the middle of the century, Alberdi and others were attempting to steer the direction of Brazilian politics towards that of its neighbors, including the United States.In 1852, Alberdi complained bitterly, "nothing is more outdated and false than the pretended antagonism between the political views between Brazil and the other South American republics [...] Brazil is today a power essentially American" (Franco and Pilla, 1958, p. 25). 8These calls came from important political fi gures.Even Rui Barbosa, one of those credited with founding a presidential Brazil, very famously became disenchanted with his creation and joined the call for parliamentarism.Nevertheless, despite legislative proposals that were presented every ten or fi fteen years (the strongest of them was Raul Pilla's amendment in 1946), a presidentialist majority (often enforced and fi nanced by clearly defi ant presidents) always prevailed. 9Parliamentarism reentered Brazilian politics by political necessity in 1961 for about 18 months.That year, vice president João Goulart, the left successor of a right administration, assumed the presidency after frustrated President Jânio Quadros stepped down.The military and the right found Goulart's accession unacceptable and began preparations for his removal.To prevent a coup, leaders from the left and right reached a compromise in which Goulart would continue as president in a parliamentary system.Stripped of nearly all power, Goulart began almost immediately to press for the return of presidentialism.After 18 months of economic and political confusion, he was able to sow enough doubt in the system that the legislature organized a national plebiscite on the question.The Brazilian mass public, as they have in subsequent surveys and in a similar plebiscite 30 years later, voted overwhelmingly for presidentialism (nearly fi ve to one).After the reinstallation of presidentialism (and with it the restoration of Goulart's power), the inevitable military coup occurred to remove Goulart, thus beginning twenty years of uninterrupted military leadership in the guise of a presidentialist democracy.and styles.The result is an interesting tension between each of these infl uences.
Another reason to focus on constitutional decisions in Brazil is that its Constitutional Assembly in 1987-1988 and the parliamentarism-presidentialism debate within the assembly marked a critical moment in the country's transition to democracy.Unlike the rather perfunctory constitutional process of some of its transitioning neighbors (e.g., Argentina), Brazil's process was open to more fundamental and comprehensive change in the structure of government.Such a deliberate, thorough process presents a good opportunity to understand how decision makers incorporate foreign models when designing new political institutions.Moreover, debate over the system of government was easily the most important and fundamental issue facing the delegates at the constitution.The subject occupied a disproportionate amount of their time, inspired over sixty books and countless articles, and its vote was the only session that all 559 delegates attended.Indeed, for some, the very reason for commissioning a new constitution at all was to rethink presidentialism.Remember that the military government had operated within a formally democratic constitution (albeit modifi ed to suit their needs in 1969).Many Brazilian politicians -at least presidentialists like Marco Maciel -argued that a new constitution was unnecessary for the transition to democratic rule.

The odds on presidentialism and parliamentarism in the Constitutional Assembly
From the day the Constitutional Assembly convened, the adoption of a parliamentary system seemed inevitable.There are four strong reasons to have expected such a choice.First, a large majority of elites within Brazil had converged on a preference for parliamentarism by the time the convention opened.Intellectuals, for one, were (and still The weight of history and the rebuilding of Brazilian democracy are) overwhelmingly in favor.Virtually all of the opinion pieces in the media and books on the subject are unequivocally supportive of parliamentarism10 .José Serra, in fact, claims in his paean to parliamentarism that he can count on one hand the intellectuals who support presidentialism.Moreover, this group of pro-parliamentarist intellectuals was well represented at the convention.Surveys of delegates11 consistently showed that more than 70% of the delegates favored parliamentarism throughout the convention.Even more importantly, the chair and rapporteurs of the relevant committees (that is, the committee and its subcommittee responsible for drafting the proposal and the integration committee responsible for incorporating additions and changes introduced by delegates in the general assembly) were staunch parliamentarists.
Second, to the extent that policy makers had formally evaluated the merits of the two systems, the results had come back strongly in favor of parliamentarism.The most important of these studies was one carried out in 1985 by a fi fty-member commission headed by Afonso Arinos.The commission, staffed largely by academics (including Bolivar Lamournier and Helio Jaguaribe), recommended a mixed parliamentary system like that of France.Fourth, the political and economic context of the 1980's seemed to predispose legislators to parliamentarism.After twenty years of military rule, marked by egregious displays of executive dominance, the stage was set for a substantial shift in power towards the legislature.Parliamentarism, many argued, was exactly the right vehicle to accomplish this delicate rebalancing.Moreover, and most importantly, parliamentarism offered an excellent solution to the problem of Executive transition during crises that had Zachary Elkins seemed to plague Brazil and its neighbors.Many Brazilian elites had lamented the infl exibility of presidentialism and its unresponsiveness to changes in the political mood.Nelson Jobim, in a critique of presidentialism, quipped that the system had only three responses to crisis: 1. suicide in 1954; 2. renouncement in 1961, and 3. coup d'etat in 1964  (apud Pereira, 1993)  12 .Parliamentarism, of course, offers a convenient and legitimate way to remove an unpopular or ineffective Executive.
Such convenience and fl exibility seemed particularly salient during the drafting of the Constitution.Many legislators were unhappy with José Sarney's succession after Tancredo Neves' sudden infi rmity and then death in 1985.One year later, when the convention opened, they were certainly amenable to a mechanism that would remove him.This discontent with Sarney was only exacerbated by hyperinfl ation and Sarney's apparent inability to resolve it.Moreover, anti-Sarney leaders were not the only ones attracted to parliamentarism as a mechanism for executive removal.So too were forces on the right who feared a successful presidential run by Lula or even Leonel Brizola.In short, parliamentarism seemed to be a good fi t at this time.
So how did Brazil wind up with presidentialism?The use of eleventh-hour carrots and sticks by a still-powerful and very presidentialist president seemed to carry the day.A parliamentary constitution sailed through the three committees only to be scuttled by a vigorous campaign by Sarney, who doled out an estimated 100 million dollars in pork in order to insure support (Fleischer,  1990).With the full assembly present -including many delegates who had largely stopped attending the plenary The weight of history and the rebuilding of Brazilian democracy sessions -presidentialism won with 60 percent of the vote.Parliamentarists were reportedly shocked at the turn of events but, with the convention coming to a close, were not able to overturn the decision.They managed to salvage some hope by incorporating the question into an already scheduled plebiscite in 1993 on the question of monarchy v. republic.An appeal to the mass public was almost certainly in vain.Throughout the century, Brazilian citizens have consistently supported presidentialism when polled or asked to vote.Sure enough, 1993 proved to be no exception as presidentialism was confi rmed by a margin of three to one.

Evidence of diffusion in the Constitutional Assembly
To what degree and in what way did the practices and experiences of other governments matter to the delegates?We look for an answer in three sources of evidence: 1. what delegates say in an interview; 2. what they argued during the Constitutional Assembly; and fi nally 3. how they voted.Together these three sources suggest the strong infl uence of foreign models.

Deliberations within the Constitutional Assembly
The best, and most illustrative, way to understand the way leaders make use of foreign experience is to read the text of the lengthy debate in the Constitutional Assembly.Of course, this gives us access only to the public discussion of the issue, and no insight into backroom deals and bargaining -a decisive arena in any political (including constitutional) issue.Nevertheless, with respect to the system of government question in Brazil, we are blessed with an extraordinary amount of public discussion on the issue.From the day of the fi rst plenary session, February 14 of 1987, up until only days before the fi nal draft was issued, September 5 of 1988, delegates debated parliamentarism Zachary Elkins and presidentialism.This discussion occurred in essentially three settings: the plenary sessions, the committee on the division of the three powers, and the powerful integration committee (the group responsible for incorporating the suggestions of the various committees into a draft of the constitution).
In the plenary session alone, there were over 350 speeches delivered on the subject.Given the vast agenda in front of the delegates, this represents a monumentally disproportionate degree of attention to this issue.Indeed, an unscientifi c comparison of my stack of photocopied systemof-government speeches with the 15 thousand or so pages of plenary session transcripts, suggests that the delegates spent at least 15% of their time discussing the issue.
I have collected information on 339 of these speeches.While I suspect that the universe of speeches is greater than 350 (but less than 400), my sample is restricted to those speeches that I was able to locate and identify as having to do primarily with the system of government.I disregard speeches that address the question only peripherally, as well as those which address merely mechanical issues of the debate such as vote calls and points of order.
Furthermore, I have sampled 80 of the 100+ speeches on the question delivered in the three Powers committee.As we may expect, the nature of these speeches, compared with those in the plenary session, are on the whole more refi ned and more substantive.These meetings convened experts (political scientists and constitutional lawyers) and commissioned a surprising amount of data and historical records on comparative systems of government.
Not surprisingly, given the parliamentarist leanings of most of the delegates, parliamentary speeches outnumber presidentialist speeches by a 2 to 1 margin in the plenary session (62% are parliamentarist, 32% are presidentialist, and 6% are unclear) and by a 9 to 1 margin in committee.This The weight of history and the rebuilding of Brazilian democracy refl ects the overwhelming preference for parliamentarism among those active and verbal during the assembly.

Salience of foreign experience to the delegates
So how relevant was foreign experience to the delegates?Of the 339 speeches in our sample, 151 (44.5%) appealed to foreign evidence in some way to make their argument (Table 3).If we consider the 80 speeches in the three Powers committee, the proportion of international arguments is even higher (67%), suggesting outside information was even more relevant to those deeply involved in the issue, and perhaps, given the committees parliamentary leanings, to those preferring parliamentarism.In a speech rather early on in the convention, on April 23 of 1987, Atila Lira anticipated the parade of foreign examples that would come before the delegates: "Much we will drink -we delegates -of the fountain of experience and wisdom of foreign politics, and from there take advantage of a valuable contribution -making, of course, adaptations which better conform with our cultural formation".
Given that most wealthy, high performing democracies are parliamentary governments, one might expect that the parliamentarists would employ more foreign references than would the presidentialists.Surprisingly, this was not the case.A roughly equal proportion of parliamentarists and presidentialists (44% and 42% respectively) cited foreign evidence in their arguments (Table 3).Upon analysis, however, it becomes clear that a signifi cant number of presidentialist examples are merely responses to the foreign references cited by parliamentarists.
It would not be right to imply that the international arguments dominated the debate.It is more accurate to say that the delegates used a pluralistic approach, using whatever evidence and logic at their disposal.Many speeches (35%) drew directly on the Brazilian experience with presidentialism and parliamentarism -a very understandable approach given country's long trials with the two systems in the last two centuries.
In fact, there was some noticeable resistance to the idea of importing practices from foreign soil: We absolutely do not want simply the transplanting of a constitutional model from another nation.What we want is a system of government suited to our political, economic, and social formation -one appropriate to the institutional reality of Brazil13 .
The weight of history and the rebuilding of Brazilian democracy I think that it matters little the name that we give the system of our government -whether it be neo-parliamentarist or neo-presidentialist.What matters is that we do not recreate the crazy experiences copied from foreign models14 .Presidentialism, then, satisfi es a national aspiration.It was not the fruit of ignorance but a conscious choice of this nation [...] Our case, cannot be one in which we copy foreign experiences 15 .
By way of introduction, it is also worth remarking upon the level of sophistication at which the debate was conducted.Both sides of the debate, but especially the parliamentarists, were well versed in the substantive and theoretical evolution and implications of each system of government.The heavy hand of the social scientist was visible throughout the debate.Delegates were not shy about introducing the ideas of Duverger, Linz, or Sartori in the plenary session, as if these authors were required reading.

Evidence of hierarchical emulation (proposition 2)
On a very basic level, proponents of parliamentarism in the Constitutional Assembly made sure that delegates knew of the correlation between modern, advanced societies and parliamentarism.Of the 201 parliamentarist speeches, 121 (62%) made this argument.Some would just cite the relationship; others would extend the logic to suggest that parliamentarism is the more evolved, modern practice.Delegates seem to sprinkle in this connection so frequently that it became a stylized fact during the convention that parliamentarism was the more "modern" system.Consider a few examples: Zachary Elkins

[...] [parliamentarism] is preferred by the civilized societies of the world today 16 ;
We know that the greatest countries of the world, with a democratic tradition, with the accumulated wisdom of centuries -as is the case in Italy, France, and England -have adopted the parliamentary system17 ; I support the implantation of the parliamentarist regime since it is molded in the experience and tradition of the countries with a highly civilized nature18 ; The total structure of the presidential system is in the process of bankruptcy.As such, the great majority of developed nations, with the exception of the United States and Finland, have already adopted the parliamentary form of government 19 ; We want to decentralize power, create an effective legislature, and modernize the country's political institutions -parliamentarism is the only way to do this 20 ; Presidentialism is the political portrait of frustrated and backward democracy the world over 21 ; It has become clear that the parliamentarist majority in this body will succeed in creating a modern system of government for Brazil 22 ; In truth, Mr. President, fellow delegates, we need a modern system like parliamentarism [...] 23 .
The connection between parliamentarism and modernity was useful in several ways to the parliamentarists.
The weight of history and the rebuilding of Brazilian democracy It enabled them to suggest that the system was a product of more developed societies, implying that the choice of the rich and successful must be superior.Societies that have produced such greatness, they reasoned, would surely have the best political institutions.However, some would extend the argument even further to suggest that parliamentarism was in part responsible for the development and success of not only their democracy, but also a host of other achievements.For example: [...] we have perceived, with clarity, that the ideal system of government -for all the countries that want to overcome underdevelopment -is parliamentarism 24 ; On the other hand, nations consumed by the war -such as Germany, France, Italy, and Japan -fortifi ed and were reborn under the parliamentary system, and transformed into world powers of greatness, in economic, cultural, and political terms 25 ; After the Second World War, parliamentarism was installed in Italy, Germany, France, Portugal, Greece -practically in all of Western Europe.If we look at those countries, we note that there is liberty, union life, participation of various parties, distribution of wealth, and an evolved and developed society 26 .
Of course, there were those on the presidential side who questioned such hopeful theses: Without a doubt, we will not increase popular participation and democracy by turning on our heels and imitating advanced industrial Europe or Japan 27 ; 24 Excerpt from a speech delivered by Sergio Spada on March 22, 1988. 25 Excerpt from a speech delivered by Joaquim Bevilacqua on April 15, 1987. 26Excerpt from a speech delivered by Egídio Ferreira Lima on March 2, 1988. 27Excerpt from a speech delivered by Florestan Fernandes on November 11, 1987.

Zachary Elkins
Germany and France are cited as developed countries of the industrialized world [...] in order to suggest that parliamentarism is superior to presidentialism.It is a matter of being part of the industrialized world.All with an elevated cultural level, free of illiteracy and with a standard of living two or three times that of Brazil.Their reasoning is the same as if we were to cite the United States to say that presidentialism was superior to parliamentarism28 .

Evidence for the infl uence of cultural peers (proposition 1)
Clearly, one of the stronger arguments of the parliamentarists is the prevalence of their system among the more advanced and developed nations.However, as some of the examples above suggest, much of this argumentation is more cultural than it is economic, with their authors implying that European practices are more appropriate for Brazil than are North American practices.In fact, 41 speeches included this argument.For example, With the proclamation of the Republic we looked, incorrectly, to the United States, a model of presidentialist government.Our cultural roots are European, and not of North American origin.Why, therefore, import an alien system, strange to our traditions, if we already had, here our own model, originated from Europe, adapted and perfected through successive administrations during the [Brazilian] empire?29 ; Why not follow, in a serious and defi nite manner, the example of countries like ours [...] most recently, a country that has the same origins as us, Spain, a country which adopted a parliamentary system of the classic form [...]30 ; The weight of history and the rebuilding of Brazilian democracy It was parliamentarism in Portugal which dislodged the military from the political process.That is our example! 31.

References to the Americas (propositions 1 and 3)
There were essentially three ways for the parliamentarists to reconcile the supposed success of United States presidentialism.One was to dismiss the United States entirely as an imperialist power not worthy of admiration and emulation.A full 16% of parliamentarist speeches included such reasoning.Often these arguments mixed an affi nity for Europe with an antipathy to the United States and the institutions it had inspired in the Americas.
[…] it has to be understood that this country needs a system of government -and this system must be new, modern.It can't be a system, for example, from Mexico or from the United States where they resolve crises by means of a revolver [...] No, no it will not be Texas which inspires us but instead the example of old Europe 32 ; [...] this institution [presidentialism] will represent yet another pact with the North American devil 33 ; […] imperialism is practiced with much more intensity in presidentialist regimes, perhaps in its most gross form, perhaps the most civilized form of United States domination 34 .
A second method, however, was to suggest the uniqueness of the United States and, accordingly, its incomparability with the Brazilian system (and that of Latin America more generally). 31Excerpt from a speech delivered by José Fogaça on July 8, 1987. 32Excerpt from a speech delivered by Walmor de Luca on July 17, 1987. 33Excerpt from a speech delivered by Victor Faccioni on July 30, 1987. 34Excerpt from a speech delivered by José Fogaça on August 8, 1987.

Zachary Elkins
In the Brazilian presidentialism, badly copied from North America, we see the centralization, the authoritarianism, the paternalism, the infl exibility, and the rigidity of the Brazilian system are accentuated and perpetuate the worst of our national politics, which we need to eliminate defi nitively in order to create a viable democracy and institutional stability 35 .Already I am very certain that the potential virtues of the North American presidential system are virtues only [in the United States] 36 .We do not have the characteristics of North American society, whose force of infl uence and participation controls the power of government 37 .
[…] we adopted presidentialism [...] in an almost literal imitation of the North American system, without attention to our economic, social, and ethnic conditions which do not compare with those of the United States 38 .
A third, and related, argument employed by the parliamentarists was to suggest that presidentialism practiced in Latin America is of inevitably a weaker -almost unworkable -variety.Those who advanced this argument, would imply that the adoption of presidentialism in Latin America incorporated the negative, more autocratic, tendencies of the system.In essence, another suggestion that Brazil is not well suited to such a system for it exacerbates the Latin American predilection towards authoritarianism."In addition, the presidentialism which spread through Latin America is an artifi cial and poor imitation of the American model.It is not built to last" 39 . 35Excerpt from a speech delivered by Victor Faccioni on April 9, 1987. 36Excerpt from a speech delivered by Nelson Aguiar on August 5, 1987 37 Excerpt from a speech delivered by Adhemar de Barros Filho on April 9, 1987 38 Excerpt from a speech delivered by Enrico Pegorano on April 9, 1987. 39Excerpt from a speech delivered by Egídio Ferreira Lima on September 3, 1987.
The weight of history and the rebuilding of Brazilian democracy Interestingly, Latin America -and its tradition of presidentialism -is used only as a negative example.Even the presidentialists do not invoke the tradition of its neighbors to support their choice of regime.
The last twenty years of military government shows the true face of presidentialism, translating what is really the typical caudilhismo of South America, in which a leader castrates the genuine path of nationality 40 .I am convinced that the presidentialist regime, in its heart, stimulates only those who nurture tyrannical and caudilhista purposes.In Europe, presidentialism no longer exists, and here in Latin America it feeds the determination of leaders of groups which have throughout the years, perpetually been in power 41 .
The presidentialists, likewise, did not use the United States as a model to any great extent.Presidentialists on the left understandably did not want to connect their institutional choice to the US.The right was much less vocal, but when pressed, would usually support their statements with evidence from Brazil.For the most part, their references were mostly reactive, trying to disabuse the assembly of the European utopia described by the parliamentarists.

Evidence of resistance to change and symbolic attachment to presidentialism (propositions 4 and 5)
Certainly, many delegates expressed anxiety about adopting an institution virtually unknown in Brazil for much of the century.João Agripino's on October 15 of 1987 statement is representative: "[...] in my opinion, if we adopt 40 Excerpt from a speech delivered by Paulo Marques on September 3, 1987. 41Excerpt from a speech delivered by Leite Chaves on April 16, 1987.

Zachary Elkins
parliamentarism in this Constitution, we will be making a leap into darkness".More than anything else, this sentiment came out in expressions of uncertainty about the foreign nature of parliamentarism.For example, "All told, I would have to say to your Excellency that, in truth, I do not share personally the idea that we ought to have presidentialism.I am a parliamentarist, but it happens that our culture will not absorb such" 42 .

Analysis of roll call votes
The argumentation used in the delegates' meetings tells part of the story.However, in the end, what mattered was their vote.Indeed, the votes -combined with what we know about the delegates -can be very enlightening.
We have argued that institutional choice often refl ects a predisposition towards certain cultural and intellectual infl uences.The Brazilian case is characterized by a unique tension between an attraction to policies of the Americas and those of Europe.Consequently, it is true that there should be some variation among delegates to the Brazilian Constitutional Assembly in how sensitive they are to each of these infl uences.If we understand the background of the delegates, then we can make predictions about their relative susceptibility to the policies from either sphere of influence (that is, the United States or Europe).If the delegates respond the way we expect given their background, then we can claim even stronger evidence about the overall infl uence of cultural assimilation in the adoption of political practices.
Accordingly, I have generated three hypotheses, each which makes a claim about a delegate's predisposition towards an American or European product.
The weight of history and the rebuilding of Brazilian democracy Hypothesis 1: Delegates with work or educational experience in Europe will favor parliamentarism; those with such experience in the United States will favor presidentialism.
This hypothesis derives directly from theory reviewed earlier about the power of contact, communication, and cultural norms.In order to test the prediction, we have gathered information on both the educational and work experience abroad by each of the delegates.For educational experience abroad, we construct two variables, US Education and European Education, for which we code delegates 1 or 0 if they received a degree in the region of interest.For work experience, we have reviewed the professional profi les of delegates and coded, again for two variables, whether or not the delegate had worked in the United States or Europe.We expect experience in either area to predispose a delegate to the policy of that area.
Hypothesis 2: Delegates from the south of Brazil will tend to favor parliamentarism.
This expectation derives from Brazil's highly diverse regional composition.Regions in Brazil are, in general, more delimited than in most countries economically, ethnically, socially, and politically.The Southern region of Brazil, a region including the states of Rio Grande do Sul, Paraná, and Santa Catarina, is largely populated by immigrants from Europe.Moreover, these areas maintain a strong attachment to Europe both emotionally and economically.Therefore, we expect that Europe will serve as a stronger reference group for delegates from the south than it would for delegates from other regions.Of course, these regions differ along important socioeconomic lines and it is essential to control for these differences in order to isolate any cultural effects (see our efforts to do this below).

Zachary Elkins
Hypothesis 3: Those in academic or law professions will favor parliamentarism; those in business professions will favor presidentialism.
This hypothesis derives from the conventional wisdom in Brazil (and much of Latin America) that business leaders are more highly connected and inspired by the United States, while lawyers and academics are similarly oriented towards Europe.A second, and intensifying, rationale for such a distinction stems from a diffusion thesis popularized by Powell and DiMaggio (1991).These authors argue that policy professionals and scholars will be more amenable to policy innovations which promise to be superior to existing policies.As we assert above, there was in Brazil at the time a rough consensus among intellectuals that parliamentarism was the superior institution.Our assumption is that academics and those in the law would be most susceptible to these opinions.

Domestic political explanations of vote choice
Of course, the alternative to these diffusion explanations is that the system of government decision was one driven by domestic political concerns.In order to control for these factors, we include a number of political predictors in the model.A fi rst, most obvious, candidate is party affi liation.In the best of times, party loyalty in Brazil is notoriously weak (Mainwaring, 1995).Party loyalties during the Constitutional Assembly were particularly fragile as a multi-party system was in its infancy after thirty years of the two party system organized by the military government 43 . 43During the fi rst year of the assembly, the majority of delegates (55%) were organized into the catch-all PMDB -the legal opposition party during the military years.A number of smaller, further left, parties accounted for another 6 or 7% of the assembly that year.Towards the end of the Constitutional Assembly, a fair number of delegates had left the PMDB to join or form smaller parties, most notably the PSDB.The right, meanwhile, was concentrated in two parties, the PFL (23% of the assembly) and the PDS (7%).
The weight of history and the rebuilding of Brazilian democracy Mainwaring and Pérez-Liñán (1997), in an authoritative analysis of the more than 1 thousand roll call votes during the assembly, fi nd very little discipline with the exception of the smaller parties on the left.Nevertheless, given the importance of the system of government question during the assembly, it is reasonable to think that parties adopted offi cial positions on at least this issue.
Furthermore, given the strong legacy of the authoritarian years, it is reasonable to expect that the loyalties defi ned by the previous two-party system would still be intact to some extent 44 .Accordingly, we include the party identifi cation of delegates during the authoritarian years, that is Arena (the military government party) and PMDB (the opposition).
Including dummy variables for the parties assumes that delegates receive cues from the party leadership.However, the party identifi cations -which can be arrayed along an ideological scale -also help us identify the ideological leaning of the delegate.As I assert above, the system of government question does not seem to have an obvious ideological identity.However, it is reasonable to assume that, like any other issue, many delegates perceived the decision based on some calculation about where the two positions fall along a left-right scale.Consequently, we use Mainwaring and Pérez-Liñán's adaptation of Maria Kinzo's ideological scale to order the delegates (by virtue of their party identifi cation) along a left-right scale 45 .
As I describe above, one of the most important infl uences on the system of government vote was the arm twisting and vote-buying of President José Sarney.Without concrete information on phone calls from the presidential palace and diversions of the federal budget, it is diffi cult to measure this crucial variable.Fortunately, one piece of information -the delegate's average absentee ratehelps us measure Sarney's infl uence.The key to Sarney's lobbying success was to lure to Brasília those delegates who had otherwise not participated (either physically or intellectually) in the assembly.These were largely rightist members who, for whatever reason, participated only marginally in the constitutional process (Fleischer, 1990).Consequently, the absentee rate makes for a very rough approximation of Sarney's infl uence.

Bivariate results
Which, if any, of these variables mean anything to the vote on the system of government?Table 4 presents the vote results for several of these groups of delegates.A star indicates a rejection (at 5%) of the hypothesis that the vote count for a category is different from the vote count at large.The fi rst block of rows shows vote differences by party.As suspected, the largest party, the PMDB, is entirely undisciplined with half the party voting for parliamentarism and half for presidentialism.On the other hand, the smaller parties on the left, and even the rightist PFL to some extent, seem to have maintained a marginal party line.These party differences appear to be independent of ideology, albeit with a faint connection between rightist groups and presidentialism.The lack of both discipline and ideological coherence on this issue is evident if we array the party results along an ideological scale, like that suggested by Kinzo (1990).Parties on either end of the spectrum are more disciplined but entirely unpredictable by their position on the scale (Figure not included).
The weight of history and the rebuilding of Brazilian democracy The weight of history and the rebuilding of Brazilian democracy As we suspected, the party lines from the military period are also meaningful.There is a twenty-point difference in the proportion voting for presidentialism between the groups divided along the party lines as they were drawn from 1964-1979.The absentee rates, our measure of the carrot and stick efforts of President José Sarney, demonstrate some fairly dramatic differences.Those with higher absentee rates -that is, those who we presume to have been recruited by Sarney -voted overwhelmingly for presidentialism.This effect, of course, is also an indicator of our complementary theory that those absent for most of the assembly were deprived of the parliamentarist pitch that predominated the assembly.
And what of our variables relating to the three diffusion hypotheses described above?There we see mixed, but hypothesis-supporting, results.Two groups of delegatesthose educated in Europe and those having missions to the US -demonstrate political predispositions based on their foreign experience.The European-educated voted for presidentialism at a comparably low 39% while those returning from missions to the US preferred presidentialism at a rate 32 points higher (71%).On the other hand, education in the US and missions to Europe do not seem to make any difference.Nor does travel within the Latin American region.
Prospects look good for Hypothesis 2 as well.Delegates from the south and north prefer parliamentarism and presidentialism, respectively, at remarkably higher numbers than do their colleagues from other regions.There is reason to think that some of this effect is due to correlated economic differences between the regions.A measure of economic development, per-capita income by state, suggests that representatives from poorer states prefer presidentialism to a greater extent than their Zachary Elkins colleagues.This difference, interestingly, parallels the choice of presidentialism by leaders of less developed states worldwide.Again, we can control for these wealth effects in a multivariate model.
Finally, consider the vote totals by profession.We see some encouraging results for Hypothesis 3 here.As predicted, business professionals tend to support presidentialism while law professionals support parliamentarism in greater numbers relative to their counterparts.Surprisingly, academics show no particular predilection for parliamentarism.I suspect part of this non-effect is the result of an overly inclusive categorization of academics which combines the scientists with the social scientists.

Independent effects
Table 5 reports estimates of the change in the probability of voting presidentialist associated with a shift in each explanatory variable -most of which are dichotomous -from their minimum to their maximum values.These estimates are generated from a logistic regression of 15 selected variables from Table 4 on the vote for presidentialism 46 .The result is that most of the effects that we witnessed in the bivariate table discussed above remain after multiple regression. 46There are 555 valid votes and three abstentions.Missing data on some of the observations brings us down to a sample size of 533.The selected variables predict 25% of the variance.Clearly then, we have omitted some predictors of vote choice, but those we have are meaningful enough.We employ the King, Tomz, Wittenberg program Clarify, to produce these estimates.The estimates from Clarify's Monte Carlo simulation routine have the advantage of incorporating both fundamental uncertainty and estimation uncertainty The weight of history and the rebuilding of Brazilian democracy Hypothesis 1 receives strong support.The effect of a European education and a foreign mission hold up in the context of controls.This is especially true of the European education, which seems to depress support for presidentialism by 36 percentage points!While we had expected multiple regression to deliver a knockout blow to Hypothesis 2, its fl agship variables -a dummy variable for the North and one for the South -remain standing with strong effects.Delegates from the South, Pseudo R-squared = .27N.B.Estimates are the change in the probability of voting for presidentialism associated with a shift in the explanatory variable from its minimum to its maximum, all other variables held at their means.Calculated from logistic regression.
the European-centric region, remain considerably less likely to vote presidentialist (a full 15 percentage points less likely).Those from North, comparably less European than the rest of the country (which serves as the reference group in the regression), are signifi cantly predisposed towards presidentialism.Crucially, the strength of these effects remains in the presence of other variables, most signifi cantly a measure of each delegate's state's wealth.
There is moderate support for Hypothesis 3. Businessmen, who we expected to use the US as a model, are comparatively presidentialist but only marginally (4 percentage points).Lawyers, who are sympathetic to arguments for parliamentarism (for both European-centric and professional reasons), are comparatively parliamentarist but also marginally (8 percentage points).
As for our controls, they largely remain moderate predictors of vote choice.Party affi liation matters a great deal for members of the PFL, PDT, and the PT.The Ex-Arena party faithful are indeed more likely to vote presidentialist (to the tune of 20 percentage points).The effect of our measure of ideology, however, seems to wash out, solidifying our impression of this issue as almost irrelevant to ideology.
The remarkably strong effect of absenteeism on the vote deserves emphasis.The delegate with the lowest attendance record was a full 44 percentage points more likely to vote for presidentialism than was the delegate with the best record.As we suggest above, there are two complementary interpretations of this effect.One is that President Sarney's campaign directed largely at absent delegates was very effective.The second is the conclusion that those who had tuned into the deliberations and the pulse of the Constitutional Assembly had adopted the studied and learned view that parliamentarism was the superior institution.
We illustrate the strength of these effects by creating some simulated probabilities for delegates of various The weight of history and the rebuilding of Brazilian democracy backgrounds (Table 6).Again, these are derived from the logistic regression estimates above.We simply fi x several variables at one value or another, keep the others at their means, and measure the model's prediction.For example, the fi rst row simulates the probability of a presidential vote for a delegate with several parliamentarist tendencies -a lawyer from the South, educated in Europe, and with no previous affi liation with the military party Arena.Given these parameters the probability of a presidentialist vote is a trifl ing 0.11.Modifying the profi le in various ways produces another set of probabilities.A delegate from the North, involved in business, with trips to the United States, with a high absentee rate is almost certain to vote presidentialist (probability of 0.98).

Interviews with delegates
Argumentation on the fl oor of the assembly and the vote behavior of the delegates provide indirect clues about how constitutional framers use foreign experience.However, the most direct method of understanding individuals' motivations is to ask them.Accordingly, I conducted interviews with 79 of the 559 delegates to the constitutional convention as well as 68 bureaucrats who had been employed in the federal government during this period.My interview method included exploratory and confi rmatory elements.On the one hand, I treated the conversations as opportunities to ask open-ended questions which would elicit singular information about the delegates' experience.So, for example, I asked respondents to describe the evolution of any bills or amendments on which they had worked, to describe the research process within their committees, and to identify attributes of foreign governments that they would like to reproduce in Brazil.
On the other hand, I also asked respondents to answer a number of closed-ended questions with stipulated response choices.Since self-determination and innovation tend to be prized over emulation and conformity, the motives I sought to uncover, I included as many experimental and unobtrusive measures in the survey instrument as possible.
We concentrate here on the responses to interview questions with particularly interesting insights about propositions presented above.One component of the survey included several very general invitations for the respondent to discuss Brazilian institutions with respect to those worldwide.While these questions were intended as mere preliminaries, they yielded some interesting results.For example, consider the question, "what is one thing that is wrong with the Brazilian constitution that should be amended?".Answers understandably ran the gamut from proposals to eliminate the constitution's many social The weight of history and the rebuilding of Brazilian democracy protections to those to reinstate the monarchy.What is of interest to me was the high proportion of responses (48%) which either compared the Brazilian constitution with those of other nations, or justifi ed a change with an appeal to the practices of another nation.This offers strong support for the saliency of foreign examples.
Upon direct questioning, however, most respondents would dismiss or downplay the infl uence of foreign models on their own decision process.I expected that some of these responses were motivated by the socially desirable bias against emulation that I mention above.By the fi fth interview, I began to vary elements of a question which asked respondents to identify their motivations for their system of government vote.In the open-ended question, "How would you say that your choice of presidentialism/ parliamentarism was affected by your feelings towards the United States and Europe", I substituted "your colleagues" for "your" for half of the respondents.While few would suggested that their own choice amounted to a choice between the institutions of the United States versus those of Europe, nearly half of those asked to speculate about their colleagues' motivations suggested as much.
Apart from the salience of foreign models, a far more basic issue concerns the level of information that decision makers possess of foreign institutions.Indeed, such knowledge is presumably a prerequisite for any actual emulation.To verify their knowledge, I asked the question, "Can you tell which countries you think of when you think of presidentialism?And when you think of parliamentarism?"The responses to the question testify to Brazilian political elites' understanding of the international distribution of systems of government.All but one of the 79 delegates named correctly at least two countries with either presidentialism or parliamentarism.
Finally, consider responses to the question, "Many times The weight of history and the rebuilding of Brazilian democracy * * * What have we learned?First, leaders are very attentive to foreign experience and foreign models when designing political institutions.Not only are their constitutional proposals largely inspired by foreign models, but these models serve as a large part of delegates' justifi cation for their position.This is despite some understandable resistance to the wholesale importation of a foreign policy, as well as a wealth of relevant domestic experience to draw on.Second, there is strong evidence that constitutional designers are highly sensitive to the infl uence of cultural and, to a lesser extent, geographic, peers.Third, it appears true that the institutions and policies of economically and politically highachieving nations tend to be especially attractive models, with the caveat that that of the United States elicits some resentment.Finally, it is clear that political institutions show some of the same evolutionary ineffi ciencies that we see in market examples typifi ed by the prevalence of the QWERTY typewriter.Namely, when institutions become the industry standard within certain peer groups, modifying or removing them is exceedingly diffi cult.
There are certainly other interesting avenues to pursue within these themes.For example, the tension in the Americas between the attraction to the United States and Europe intriguing.Historically, the United States served as a compelling, young, irreverent, and vibrant model for democracies evolving in the 1800's.Today, likely because of the substitution of the United States for Europe as the "imperial" power as well as an increasingly poor fi t between the United States political structure and that of Latin America, the European model appears more relevant.This role reversal is worthy of more attention.
What are the implications of this study?There is reason to believe the diffusion properties we describe in the Brazilian

Table 1 :
Characteristics of Presidential and Parliamentary GovernmentsSource: World Bank;Alvarez et al. 1999

Table 3 : Attributes and Arguments in System of Government Speeches Plenary
Session, Brazilian Constitutional Assembly 1987-1988

Table 6 : Simulated Probabilities of Voting for Presidentialism for Exemplar Delegates Profi le Probability of Voting for Presidentialism 95% Confi dence Interval
N.B.Estimates calculated from logistic regression.All other variables held at their means.