1 Introduction

While in its initial phases of development constitutional political economy was primarily “a normative endeavour interested in legitimising the state and its actions in terms of fairness, justice, or efficiency”, then by now the research programme has clearly expanded and includes an increasing number of positive analyses (Voigt and Wagener 2002, p. 15). Following Buchanan (1990), the research programme of positive constitutional political economy focuses on two main questions: how are constitutional rules chosen (constitution as explanandum) and how do they influence policy choices and outcomes (constitution as explanans)? (Voigt 1997; Voigt and Wagener 2002).

The explanandum stream of constitutional political economy—which is the focus in this paper—analyses how different factors in constitution-making influence the resulting content of the constitution (Voigt 1999, p. 2). Although the explanandum stream is still at an early stage of development, it has received increasing attention (owing to EU constitution-making, among other factors) in recent years. Of particular interest have been the questions of how the process of constitution-making, the decision-making procedures employed by the constituent body, the constellation of actors involved, their preferences and bargaining power, the models they can (or are expected) to draw on, time-limits and other constraints influence the content of the constitution. While all these factors are likely to influence the resulting document, this paper has chosen one particular aspect for closer examination: it is concerned with the effects of the kinds of participants involved in constitution-making on the content of constitutional provisions proposed during constitution-making, with specific attention to the provisions pertaining to legislative, executive and judicial powers.

The goal of this paper is to contribute to the literature in positive constitutional political economy by bringing together the hypotheses that predict the constitutional preferences of particular kinds of participants (with regard to the delineation of powers of different government branches) and by testing them with two specific cases: the Constitutional Assembly of Estonia (1991–1992) and the Federal Convention (1787). In particular, this paper focuses on whether a framer’s institutional position, partisan background and constitutional expertise provide sufficient grounds for predicting his or her constitutional preferences.

The case of constitution-making in Estonia offers a remarkably useful testing ground for constitutional political economy for a number of reasons. Unlike in most other Central and Eastern European (CEE) countries, where the post-1980 constitutions were drafted by existing legislatures, the 1992 Constitution of Estonia was prepared by a separate constituent body. As the Estonian Constitutional Assembly had two kinds of members—those who were members of the existing legislature and those who were not—it allows one to examine whether the delegates from the existing legislature had different constitutional preferences than other delegates. Furthermore, as a number of local and foreign constitutional experts were involved in drafting the Estonian Constitution—alongside with the members of the Constitutional Assembly—the case offers an opportunity to analyze whether the constitutional preferences of practising politicians and constitutional experts diverged. Taken together, such a constellation of different kinds of framers in Estonia provides a special opportunity for testing the hypotheses concerning the constitutional preferences of different types of framers.

While the instances of constitution-making in most other CEE countries have been examined in the light of the hypotheses proceeding assuming framers’ institutional and partisan interests (see Elster 1993, 1995, 1996a, 1997b; Adamovich 2004), the case of constitution-making in Estonia has not been considered through that lens so far. The hypotheses concerning the constitutional preferences of constitutional experts have not been explicitly tested by any studies so far.

The main sources of information for the case study on Estonia were the verbatim records of the plenary sessions of the Constitutional Assembly (i.e., word-to-word recordings of what was said at the discussions of the Assembly, published in Põhiseadus ja Põhiseaduse Assamblee 1997), the memoirs of the members of the Assembly, archival materials and secondary materials describing and analyzing constitution-making in Estonia.

In order to put the findings of the case study on Estonia in a broader perspective, the results are juxtaposed with the “paradigmatic” case of modern constitution-making—the Federal Convention of 1787. Since there already is a vast literature exploring the US case, this paper does not aspire to be a comparative study in a conventional sense. The “primary” case study is Estonia, while the US case serves as a “secondary” case (see Levi-Faur 2006 for a discussion on this distinction) for providing further tests of the hypotheses, illuminating contrasting outcomes and drawing attention to features of constitution-making that appear to be the same, regardless of the fact that these two instances of constitution-making took place in very different time periods and in very different contexts.

The paper is organised as follows. The second section presents an overview of the theoretical discussion and hypotheses pertaining to the effects of the kinds of participants involved in constitution-making on the proposed content of the constitution. The third section gives an overview of constitution-making in Estonia in 1991–1992, and the fourth section tests the hypotheses outlined in the second section with the Estonian case. The fifth section discusses the findings of the Estonian case in the light of the constitution-making at the Federal Convention of 1787; the sixth section concludes.

2 Framers’ constitutional preferences: overview of the theoretical discussion

The existing theoretical and empirical works in positive constitutional political economy have put forth a number of hypotheses concerning the effects of the kinds of actors involved in constitution-making (and their motives) on the proposed constitutional provisions. In their analysis of constitutional preferences, Vanberg and Buchanan (1989) distinguish between constitutional theories and constitutional interests. A person’s constitutional theories are his or her predictions about the prospective outcomes of alternative rules; constitutional interests are the person’s subjective evaluations of the expected outcomes of the rules.

Constitutional interests, in turn, can be carried by different kinds of motivations. Elster (1993, 1995, 1996b, 1997a, b) distinguishes between personal, group and institutional interests in constitution-making. As Elster (1993, p. 181) explains, personal interest refers to the “purely personal interest of the founders in a constitution that favours them economically or otherwise”.Footnote 1 Group interests refer to the interests of the constituencies (be it supporters of a party or inhabitants of a territorial sub-unit) that a framer represents (Elster 1995, p. 378, 1997a, p. 132). Institutional interests refer to the interests of framers to advance the constitutional powers of the institution they are members of (be it legislature, executive or judiciary).

Vanberg and Buchanan (1989) underline that in constitutional choices, constitutional theories usually play a more important role than in ordinary choice situations. In contrast, Elster points to the potentially dominating role of framers’ interests in constitution-making. Although Elster (1995, p. 377) notes that the personal interest of framers in specific constitutional provisions is in most cases relatively marginal, institutional and group interests can play a major role in constitution-making. The extent to which constitutional interests come to the fore in constitution-making depends on the nature and set-up of the constitution-making body.

As Elster (1993, p. 182) puts it, “The interest of political institutions appears most clearly when the institutions to be regulated by the constitution also take part in the constitution-making process.” Hence, Elster (1993, 1995, 1996b, 1997a, b) argues that when the constitution-making body coincides with the existing legislature (or is a sub-committee thereof) and/or other branches of government are involved, institutional self-interest would play an important role in constitution-making: the framers would be inclined to enhance their own position in the constitutional framework, at the expense of other organs of state. As Elster (1996a) explains, institutional interests can reflect different kinds of motivations, like the intention to stay in political life beyond the constituent parliament or identification with the institution that the actor belongs to (e.g., because of the need for cognitive consonance). Proceeding from the assumption of framers’ institutional self-interest, Elster (1995, 1996a, 1997b) hypothesizes that constituent assemblies that simultaneously serve as ordinary legislatures will assign more extensive constitutional powers to legislature than to executive and judiciary. When predicting the preferences of individual framers, one would hence expect those framers who come from an existing legislature to argue for strong constitutional powers of legislature. Elster (1995, 1996a, 1997b) also conjectures that if the incumbent president is involved in the constitution-making process, he will promote a strong presidency.

Group interests can be expected to be relevant in constitution-making when the framers know that their constituents will be involved in ratifying the final document (Elster 1993, 1995). When framers represent particular territorial subunits of the country and these subunits have to ratify the constitution, the framers will find it difficult to ignore the territorial interests when drafting the constitution. To the extent that the constituent body is made up of groupings clearly identified with particular parties, partisan interests are likely to play a role in constitution-making, alongside with institutional interests. Voigt (2004), Mueller (1996, ch. 21), Elster (1993) conjecture that the involved political parties will try to influence the constitution—in particular the electoral rules—in a way that enhances their chances of being (re)elected. Thus, small parties will favour proportional representation with a low threshold, while large parties will endorse a plurality (or majoritarian) system. Politicians who are already well-known and popular will endeavour to establish rules that allow voters to cast a ballot for specific individuals, while disliked leaders of the main parties will advocate party-list procedures. Elster (1993, p. 182) also suggests that a party that has a strong presidential candidate will argue for a powerful presidency, whereas other parties will seek to limit the constitutional powers of the president. Further, the party that has a strong presidential candidate will argue for an election mode that will enhance the likelihood of that candidate attaining that office (Elster 1997b, p. 233). Similar hypotheses, reflecting the assumptions of institutional and partisan interests, as put forth by Elster, Voigt and Mueller, have been suggested by Geddes (1996), Magalhaes (1999), Smithey and Ishiyama (2000), Ginsburg (2002), Frye (1997).

In addition to the representatives from existing governing institutions, external constitutional experts can be involved in drafting the constitution.Footnote 2 The analytical value of looking at the role played by constitutional experts has been emphasised by Elgie and Zielonka (2001, p. 39), Elster (1995, pp. 395–396), among others. Voigt (2004, p. 217) conjectures that constitutional experts who are involved in constitution-making will not attribute a dominant role to the legislature vis-à-vis the other two branches. In addition, he hypothesises that if the experts involved are constitutional lawyers, they will advocate extensive guarantees of independence for the judiciary.

Taken together, the main hypotheses that are tested in this paper are the following:

H1::

Members of constitutional assembly who come from existing legislature will seek to give overriding importance to the legislative branch, at the expense of the executive and the judiciary

H2::

If the incumbent president is involved in constitution-making process, he will promote a strong presidency

H3::

Members of constituent assembly who are members of a party support constitutional rules that foster the electoral chances of their party in legislative elections; supporters of a particular presidential candidate support electoral rules that enhance the likelihood of that candidate becoming a president

H4::

Constitutional experts who are involved in constitution-making will not attribute a dominant role to the legislature vis-à-vis the other two branches

H5::

Constitutional experts who are constitutional lawyers will advocate extensive powers and strong guarantees of independence for the judiciary

3 Overview of the constitution-making in Estonia, 1991–1992

During the coup in Moscow on 20 August 1991, the Supreme Council of Estonia declared the independence of Estonia and decided to convene the Constitutional Assembly (CA), which would include delegates from two competing representative bodies in Estonia at the time: the Supreme Council and the Estonian Congress. The Supreme Council (105 members) had been elected in the first multiparty parliamentary elections in March 1990 by the electorate that consisted of all inhabitants who were registered in Estonia. The Estonian Congress (499 members) had been elected in February 1990 by all who had been citizens of Estonia prior to the Soviet occupation (and their descendants). The elections to the two bodies reflected the rivalry between two different pro-independence groups that had been competing against each other since 1989. On the one hand, the Estonian Popular Front favoured a more flexible approach to independence: making use of the existing institutions, seceding from the Soviet Union and proclaiming a new Estonian Republic. On the other hand, the movement called the Citizens’ Committees (promoted by the Estonian Heritage Society, the Estonian National Independence Party and the Christian Union) opposed such a conciliatory strategy and working within the existing “illegitimate” institutions; they demanded establishing new institutions, called for the establishment of the Estonian Congress and sought the restitution of the pre-war Estonian Republic. The Popular Front participated in elections for the Supreme Council in 1990 and won a plurality of seats. The National Independence Party, together with the Estonian Heritage Society, became a dominant force in the Estonian Congress. While the Supreme Council became the official legislature, with the authority to enact laws, the Estonian Congress evolved into a “shadow parliament” that deliberated on issues of constitutional nature and criticised the Council. (Pettai 2001, pp. 112–114; Taagepera 1993, ch. 7, 1994, pp. 213–214; Raitviir 1996, p. 353; Lauristin and Vihalemm 1997, p. 90).

The decision of the Supreme Council to convene the Constitutional Assembly has been termed a “miracle compromise” (Pettai 2001, p. 111), given that it managed to strike a balance between the diverging plans of the Supreme Council and the Estonian Congress with regard to how to proceed with constitution-making. The resolution on the election of the Constitutional Assembly (3 September 1991) foresaw that the Constitutional Assembly would have sixty members (thirty from the Estonian Congress and thirty from the Supreme Council); the election of the members to the CA was to follow the principle of proportionality, but the exact procedure was left for the bodies themselves to decide. The Assembly was given the right to contract experts and scientists for advice. The draft Constitution was to be submitted to the Supreme Council (by 15 November 1991), which would then decide on putting the draft on referendum. Taagepera (1994, p. 217) notes that the election outcomes in both bodies were fairly proportional to the strength of the various groupings in the respective bodies. Among the total of sixty CA members, about twenty were connected to the Popular Front, another twenty could be counted as national radicals, thirteen were various Estonian moderates and reform Communists, and seven were representatives of the Russians. Among thirty representatives elected from the Supreme Council, fifteen were also members of the Estonian Congress; among thirty representatives elected from the Estonian Congress, three were also members of the Supreme Council (Aaskivi and Pärnaste 2002, p. 49).

In order to facilitate a more detailed discussion on different chapters of the Constitution, the rules of procedure foresaw the creation of committees. At the first meeting of the Constitutional Assembly, on 13 September 1991, seven topical committees were formed. By mid-October, five different draft constitutions had been submitted by various groupings (from the Assembly and the incumbent government) and the CA voted on 11 October 1991, which of the drafts would be taken as the basis for continued work. The main competing drafts were the draft of Jüri Raidla (the Minister of Justice) and his team (consisting of leading constitutional experts in Estonia) and the draft of Jüri Adams (a member of the Estonian Congress and the chairman of the Estonian National Independence Party).Footnote 3 The draft of Raidla put forth a hybrid system with a strong president, while the draft of Adams entailed a parliamentary system with a weak president. The Constitutional Assembly chose the draft of Adams as the basis for further deliberations (with 29 votes over 22).

At the end of October 1991, a delegation of foreign experts visited the Assembly and commented on the draft.Footnote 4 The Constitutional Assembly did not manage to meet the initial deadline (set for November 15) and asked for extending the deadline for the submission of the final draft. The Supreme Council decided that the CA must present the draft Constitution by 20 January 1992. The first draft of the Constitution was released to the public on 21 December 1991. About 500 persons, groups and organisations sent their recommendations and some of the proposals were incorporated in the new version of the draft. In January 1992, the Supreme Council decided that the Constitutional Assembly should continue its work but cooperate more closely with the “expert group” led by Raidla (Aaskivi and Pärnaste 2002, p. 60). The expert group consisted of seven members; all of them had been members of the team that had formulated Raidla’s draft.

A next version of the draft Constitution was approved by the Constitutional Assembly on 14 February 1992 (by a margin of 32 to 3, with 6 abstentions) and it was submitted to the Supreme Council for consideration. In March, the Supreme Council returned the draft to the Assembly, claiming that it lacked consensus among the Estonian people (especially with regard to the mode of electing the president) and asked the CA again to “coordinate” its draft with the local “expert group”, to take into account the proposals of the Supreme Council and to submit a new draft by mid-April. (Taagepera 1994, pp. 226–227; Aaskivi and Pärnaste 2002, p. 68; Adams 2002, p. 151; Hänni 2002, p. 166) During the following month, a host of minor wording changes were made to the Assembly’s draft, but in terms of content, not much was changed. A major substantive change was the right given to the president to appeal to the Supreme Court when his veto is overridden by the parliament. The Assembly reaffirmed its support for the document at the thirtieth session, on 10 April, and the Supreme Council decided to put the Constitution on referendum. On 28 June 1992, the Constitution was approved in a referendum where only Estonian citizens could vote. The list of citizens included 669,080 persons; 446,708 took part in the referendum, of whom 91 percent voted in favour of the Constitution.

4 Testing the hypotheses: the Estonian case

The goal of this section is to test the hypotheses outlined in Sect. 2 with the case of constitution-making in Estonia. As mentioned above, since the Estonian Constitutional Assembly had two kinds of members—those who were members of the existing legislature and those who were not—it allows one to examine whether the delegates from the existing legislature (the Supreme Council) had different constitutional preferences than those who came from the Estonian Congress (which was a representative body for deliberating on issues of constitutional nature but did not have the authority to enact laws). Further, as a number of constitutional experts (both foreign and local) were involved in drafting the Estonian Constitution, the case lends itself to testing whether the constitutional preferences of the experts and the active politicians diverged.

With regard to the relative powers of the legislature and the president, the Estonian case confirms hypothesis H1—at least at first sight.Footnote 5 More than half of the framers came from an existing legislature (the Supreme Council) and the resulting Constitution is parliamentarian (with the president as a symbolic head of the state with limited executive powers). The Constitution gives the residual powers to the legislature, and as Frye (1997) notes, the presidency in Estonia is among the weakest in the region. Schneider (1999, p. 419) even argues that the Estonian Constitution gives the legislature several tasks that are untypical functions for a parliament and are usually attributed to the executive.

In order to provide a more accurate test of H1, however, one should examine the constitutional preferences on individual level as well. Of special theoretical interest is whether extensive powers of the parliament were primarily favoured by delegates from the Supreme Council or by delegates from the Estonian Congress. In terms of proportions, more representatives from the Estonian Congress than from the Council preferred a strong parliament and a weak president. In fact, the Supreme Council (acting as an upstream authority) repeatedly sent the draft Constitution back to the CA and sought to persuade the Assembly to strengthen the powers of the president (alongside with the requirement to replace indirect elections of the president with direct popular elections). In other words, the existing legislative body as a whole appeared to prefer a stronger president (and a weaker parliament) than the CA itself and the most vocal proponents of a strong presidency came from the existing legislature.Footnote 6 Altogether, when considering hypothesis H1 as positing an effect-of-a-cause prediction with regard to the relative powers of the legislature and the executive—stating that a framer who is a member of an existing legislature will support a strong legislature and a weak president—the case of Estonia provides only partial evidence for it. Although a majority of the framers in the CA who came from the Supreme Council favoured a strong legislature, a significant number of Supreme Council delegates did not. The fact that a majority of the existing legislature as a whole supported a constrained legislature and a strong president weakens the evidence for H1 even further.

When one wants to make a more demanding use of H1 and consider it as offering a cause-of-an-effect explanation—pointing to the prevalence of framers’ institutional interests in constitutional choices—the Estonian case calls for even more caution. Although Elster (1995, p. 380) claims that institutional interest was the most important factor in the making of the post-1980 constitutions in Eastern Europe, the Estonian case shows that it is problematic to attribute the constitutional choice of a strong legislature and a weak president to the fact that a majority of the framers came from the existing legislature. On the one hand, indeed, even in the constitutional preferences of the delegates from the Congress, institutional interests played some role, as they were prominent figures in the Estonian political scene and could consider themselves to be very likely members of the future parliament. On the other hand, it appears that among those Estonian framers whose preferences with regard to the powers of the president were primarily influenced by interests (rather than theories), it was partisan interest—rather than institutional interest—that played a major role. A number of CA members (both from the Congress and the Council) have admitted that they did not want to design a strong institution of the president, since they did not like Arnold Rüütel, the most likely candidate for that position, and the party he was affiliated with (see e.g., Kask 2002, p. 180; Lang 2002, p. 219; Hänni 2002, p. 164).

Furthermore, attributing the choice of a strong legislature merely to framers’ constitutional interests would present a rather distorted version of the story. Alongside interests, the choice of a strong parliament in Estonia was inspired by constitutional theories referring to problems that a strong president may pose for a young fragile democracy. In pointing to such constitutional theories, the framers employed examples from political science literature (especially from the works of Matthew Shugart, who was also an advisory expert to the CA) and from the experiences that Estonia had had with different constitutions in the inter-war period. Schneider (1999) suggests that the “excessive burdening” of the parliament in Estonia reflected the historical experiences: on the one hand, there was the memory of the authoritarian “silent epoch” in the 1930s, associated with a strong president; on the other hand, there was the reminiscence of the super-democratic 1920 Constitution, which had acquired the nature of a “myth” and given rise to the attitude among a number of the members of the CA that the more tasks the parliament has, the more parliamentarian the state is. In addition, like in the rest of the CEE region, the constitutional debates in Estonia were influenced by what Malova and Haughton (2002, p. 103) call “the European inheritance of parliamentarism”, which was buttressed by the Estonian framers’ will to “return to Europe”. Another factor that speaks against the overwhelming influence of framers’ institutional self-interests on the constitutional proposals in the Estonian case is the fact that a number of delegates from the Congress, who, in principle, could have been motivated by the expected membership in the future parliament, supported re-enacting the 1938 Constitution, which accorded extensive powers to the president. The main reason behind such a commitment to this constitution was the desire to signal the restoration of the inter-war Estonian Republic; the re-enactment of the 1938 Constitution would have reaffirmed the legal continuity of the state. Thus, although in the light of the theoretical discussion presented in Sect. 2, one is tempted to attribute the extensive powers given to the parliament in Estonia to institutional interests of the framers, the underlying mechanisms behind that constitutional choice were clearly more complex.

That part of H1 that concerns the design of the judiciary is clearly refuted by the Estonian case. Despite the prevalence of delegates from the existing legislature in the Constitutional Assembly, the judiciary was given extensive constitutional powers and guarantees of autonomy. Salzberger and Voigt (2002, p. 41) note that among eight CEE countries they studied, the Estonian constitutional court has the highest de jure independence. Furthermore, Estonia is the only country in CEE where judges are guaranteed life tenure by the constitution. This provision was included already in the drafts of Raidla and Adams and was accepted by members of the CA; only once was a concern expressed with regard to the applicability of this provision in the immediate post-constitutional phase because of a lack of well-qualified and professional judges (Põhiseadus ja Põhiseaduse Assamblee 1997, pp. 207–208). The general attitude among framers—irrespective of whether they came from the existing legislature or from the Estonian Congress—was that guaranteeing the independence and autonomy of judiciary was extremely important for establishing a genuinely democratic order and avoiding the politicisation of courts, which had characterised the Soviet period. Most delegates agreed that the politicisation of courts had been responsible for many tragic experiences in the Soviet period (Põhiseadus ja Põhiseaduse Assamblee 1997, p. 445). With regard to constitutional review, Pettai (2005, p. 93) notes that during the discussions of the CA, “the need for some kind of constitutional review was never disputed”, and the debates were confined to the specific “mechanics” of how the procedures and institutions would look like. The discussions on constitutional review in the CA appear to corroborate the arguments by Shapiro (1999, p. 218), who has noted that constitutional review has been adopted by framers in many countries partly because it has become an obligatory component of prevailing conception of constitutionalism. The only provision that was contested with regard to constitutional review in Estonia concerned the right of the president to turn to the Supreme Court—after the law the president has sent back to the parliament is passed unamended—and ask the Court to declare the law unconstitutional. Although left out of the early drafts, this right of the president was included in the final version. Thus, with regard to the constitutional design of the judiciary, the Estonian case provides contrary evidence to H1, irrespective of whether one treats it as an effect-of-a-cause prediction or a cause-of-an-effect explanation. Altogether, strategic considerations with regard to the judiciary played a significantly lesser role compared to the “constitutional theories” pointing to the need to break with the Soviet heritage and to establish a European-type judicial system.

Hypothesis H2 is supported by the Estonian case. Although Estonia did not have an institution of president during the time of constitution-making, there was a position—the Chairman of the Presidium of the Supreme Council—that was in many ways perceived as an equivalent to that of president or head of state. Arnold Rüütel, who held this position at the time, was also a member of the Constitutional Assembly but could not participate in its work very actively, given his numerous obligations elsewhere. Still, through his position as the head of the Supreme Council, Rüütel could influence the reactions of the Council to the draft constitutions; supported by several groupings in the Council who were interested in guaranteeing that Rüütel would become the president, the Council repeatedly sent the draft constitution back to the Assembly and called for changing the provisions pertaining to the institution of president (Laar 2002; Ruutsoo 2002a, b). The expert group led by a member of the executive (Raidla) also proposed a strong presidency. However, since Raidla’s draft was rejected and Rüütel’s affiliates were not a dominant force in the Constitutional Assembly, they could not “translate” their constitutional interests into actual constitutional choices.

The Estonian case provides confirmation for hypothesis H3. None of the parties involved in the CA had an overwhelming majority and there were no proposals to introduce a majoritarian system for parliamentary elections. It appeared to be in the interest of all parties to have a proportional electoral system. With regard to the mode of electing the president, those who supported Rüütel advocated direct popular elections (given Rüütel’s popularity among the public), while his opponents supported electing the president by the parliament (which implied at least some possibility of Rüütel not becoming a president). Since the opponents of Rüütel prevailed in the CA, the Constitution foresees that the president is elected by the legislature—despite the repeated attempts of the Supreme Council to impose popular elections of the president. Here, the CA also went against the prevailing public opinion; according to opinion polls, over 75 percent of the population supported popular presidential elections (Pettai 2001, p. 124). Thus, the Estonian case demonstrates that partisan interests may also diverge from constituents’ interests (at least as perceived by the constituents themselves). Still, in order to appease the general public (and also the Supreme Council), the implementation chapter of the Constitution foresaw that the very first presidential elections would be popular, at least in the first round, with the parliament stepping in only if none of the candidates received more than 50 percent of the vote. Curiously, while the supporters of Rüütel hoped that such an arrangement would still secure him the post, his opponents expected such higher threshold to prevent that. As it turned out in September 1992, the opponents of Rüütel had been correct.

With regard to hypotheses H4 and H5, the Estonian case provides more evidence in favour than against. The case study indicates, however, that one should distinguish between different kinds of constitutional experts and between various ways experts can be involved in constitution-making. The experts can be elected as members of the Assembly, submit draft constitutions, be contracted as advisors to the assembly, be involved through special sessions where the experts present their opinions, and the constituent body can even be compelled to work with a task-force of experts by the upstream authority. When analyzing the role of experts in constitution-making in Estonia, first, a distinction should be made between foreign and local experts. Second, among the local experts, one should differentiate between those who were contracted by the Assembly itself and a group of legal experts with whom the Assembly was, in a later phase of its work, “compelled” to cooperate (by the Supreme Council).

The local constitutional experts (who were, for most part, constitutional lawyers) who had written one of the draft Constitutions and who were “compelled” to work with the Assembly endeavoured to constrain the powers of the legislature and to increase the powers of the president and the judiciary. On the whole, the constitutional experts of Estonia were very fond of the 1938 Constitution and the way it divided power between parliament, government and president; they saw the president as having an important “balancing role”. Also, the local constitutional experts sought and succeeded to give the president the right to turn to the Supreme Court when his suspensive veto is overridden by the parliament, which increased the powers of both the president and the Supreme Court.

The foreign experts (visiting the Assembly under the auspices of the Council of Europe), who were for most part constitutional lawyers, advocated a parliamentary system and warned against a too strong president. Further, some foreign experts emphasised the need to strengthen the powers of the parliament vis-à-vis the cabinet. Matthew Russell (from Ireland), for example, pointed out that the parliament should have a stronger position vis-à-vis the government in asking questions and information. He suggested that any individual member of the parliament should have the right to ask questions from the government. (Põhiseadus ja Põhiseaduse Assamblee 1997, p. 325) Only Guy Carcassone (from France) was concerned with the weakness of the executive (p. 324). All foreign experts favoured strong and independent courts with extensive powers of constitutional review. Interestingly, some foreign experts even considered the Estonian Constitution to give too strong guarantees to the judiciary; expert Herman Schwartz (from American University, Washington), for example, pointed out that life tenure for judges was too long.

Among the experts who were contracted by the committees of the CA to give specific advice, legal specialists dominated. There were very few political scientists—with a notable exception of Matthew Shugart. The experts who advised the committees had varying opinions about how strong the different branches should be. Shugart and his co-author Rein Taagepera, another world-class political scientist (also a member of the CA), for example, advocated a strong parliament and a constrained president.

Summing up, the preferences of the local experts who were constitutional lawyers confirm the predictions made by hypotheses H4 and H5: they were suspicious of giving too many powers to the parliament and suggested increasing the powers of the president. The foreign experts, in contrast, supported a strong parliament and warned against a strong president. This indicates that the preferences of constitutional experts are not necessarily biased against a strong legislature, as H4 predicts. With regard to the autonomy and powers of judiciary, most local and foreign experts emphasised the importance of strong and independent judiciary and supported granting the Supreme Court extensive powers of constitutional review—hence corroborating the predictions made by H4 and H5.

5 Discussion: the constitutional assembly of Estonia and the federal convention compared

In this section, the findings of the Estonian case study are juxtaposed with the “paradigmatic” case of modern constitution-making—the Federal Convention (1787), which drafted the Constitution of the United States. The comparison serves, on the one hand, to provide additional tests of the hypotheses outlined in Sect. 2; on the other hand, the discussion brings to the fore issues that were similar in these two very different settings, and which would hence call for further theoretical discussions in constitutional political economy.

As the Estonian case demonstrates, even if framers come from an existing legislature, not all of them prefer a strong legislature and a weak president and, even if a strong legislature is opted for, that may be inspired by other motivations alongside framers’ institutional interests. The shortcomings of H1 are endorsed by the US case. According to H1, one would expect those delegates in Philadelphia who had served in the Confederation Congress to argue for a strong legislature.Footnote 7 Almost three-fourths of the delegates in the Federal Convention had served in Confederation Congress; nine, like Madison, simultaneously held seats in both bodies during the summer of 1787 (Stewart 2007, pp. 25–41). With regard to constitutional preferences of these framers, though most of them agreed that the national legislature should be strengthened, they had widely diverging views on how far this should go. Having been a member of the Confederation Congress did not automatically imply that a framer supported extensive powers of the national legislature. As the delegates in the Confederation Congress had been representatives of their respective states, a number of delegates sent to Philadelphia (e.g., from Connecticut and New York) continued in this role and preferred to keep the powers of the national legislature constrained vis-à-vis the state legislatures (see Stewart 2007, p. 90). On the other hand, there were a number of founding fathers who had been deeply frustrated with the weak and dysfunctional Confederation Congress and therefore supported a stronger national legislature: a body that would be able to deal with the challenges that the Confederation faced (Rakove 1997, p. 38; Stewart 2007, p. 25). Madison’s experiences as a member of the Confederation Congress, for example, certainly contributed to his ideas of a stronger national government (including a strong national legislature), which found expression in the Virginia plan (Rakove 1997, p. 37). Thus, the constitutional preferences concerning the powers of the national legislature were, inter alia, contingent on whether a framer was “a small republic man” or “a large republic man” (Jillson and Eubanks 1984).Footnote 8 Still, even “a nationalist” like Madison was skeptical of unbridled powers of legislature, which is why his constitutional proposals foresaw a joint executive-judicial council of revision, armed with some veto powers over national legislation (Rakove 1997, p. 53). Also, the four framers who most vigorously supported an energetic executive (Alexander Hamilton, James Wilson, Gouverneur Morris and Robert Morris) were those who had been members of the Confederation Congress (Rakove 1997, p. 255).Footnote 9

In sum, the US case provides only partial evidence for H1: although most of the delegates who had been serving in the national representative body supported strengthening the national legislature in the new constitution, they had widely diverging preferences with regard to how far that should go. Furthermore, the problems of legislative excesses that the framers had experienced in state legislatures also inspired them to support stronger checks and balances between the various powers on the national level.

What makes examining the effects of framers’ previous legislative experiences on their constitutional preferences even more complicated in the US case is the fact that the preferences of the delegates shifted over the course of the Convention’s work (for detailed discussions on the role of sequential decision-making in Philadelphia, see Slez and Martin 2007; Heckathorn and Maser 1987; Jillson and Eubanks 1984). For example, a number of representatives of small states (especially from New Jersey and Delaware) started supporting stronger powers of legislature after equal representation in the Senate had been secured by the Great Compromise (Mitchell and Mitchell 1964, p. 59; Lynch 1999, pp. 18–19). Conversely, after the Great Compromise, a number of delegates from larger states (Virginia being the prime example) who had argued for broad powers of Congress started striving for specific enumeration of congressional powers (Lynch 1999, p. 9). Also, several framers who had initially supported a dominant Congress (and given less consideration to how the executive should be designed) started arguing for a stronger executive after the Great Compromise. They feared that equal representation in the Senate, with senators elected by state legislatures, may make this body overly parochial and less dedicated to the “national” cause (Rakove 1997; Lynch 1999). Thus, when making predictions about framers’ constitutional preferences with regard to the powers of different branches, one has to account for the fact that these preferences can be strongly contingent on the mode of representation foreseen for composing these bodies.

In the constitutional choices concerning the legislature and the executive, the experiences gathered with previous constitutional arrangements played an important role in both cases. In the Estonian case, the problems of “ultra-democracy” that the 1920 Constitution gave rise to (including extreme fractionalisation of the legislature and short duration of governments) had some parallels with the experiences the framers in the United States had had with the excesses of the unbridled state legislatures (Lynch 1999, p. 25). Among the Estonian framers, the dangers of a strong presidency, associated with the 1933 constitutional amendments (followed by an autocratic government) and the 1938 Constitution had some parallels with the scepticism that some of the framers in Philadelphia had about a strong executive after the experience with the British monarchy (see, for example, Mitchell and Mitchell 1964, p. 77; Rakove 1997, p. 253). Thus, in many ways, both assemblies sought to find a balance between these two “extreme” forms of government.

As mentioned above, H2 finds support in the Estonian case. Although no institution of president existed at the time, the holder of a similar position (and the most likely candidate for the presidency to be established), supported creating a strong presidency. When comparing the Estonian case with that of the Federal Convention, a number of parallels and contrasts emerge. Like in Estonia in 1991, in summer 1787, there was no existing institution of president yet. At the Federal Convention, too, the framers were sure who the first president would be. The difference is, however, that though George Washington, the most likely first president, chaired the Convention, he did not actively promote a strong executive and though he consented to his expected position as president, he did so with some reluctance (Mitchell and Mitchell 1964, p. 32). Thus, H2 does not find corroboration in the US case. Another difference between the cases was that while in Philadelphia, Washington’s candidacy for that position was almost unanimously supported (Lynch 1999; Mitchell and Mitchell 1964; Stewart 2007), in Estonia, the supporters of the most likely president formed only a minority in the Constitutional Assembly. While the popularity of Washington made the framers less reluctant to design a (potentially) strong executive (Rakove 1997, p. 244), the prospect of having Rüütel as the president made the members of the CA in Estonia hesitant to provide for strong presidential powers. In both cases, drafting the powers of the president was made especially challenging by the fact that the decision on the mode of electing the president was changed a number of times.

With regard to hypothesis H3 and the mode of electing the president, further noteworthy parallels arise from the comparison of the two cases. Despite the fact that the Estonian Constitutional Assembly was convened more than two centuries later than the Philadelphia Convention and in a completely different setting, the issue of how to elect the president was one of the most contentious questions in both constituent bodies. In both cases, there were several votes on electing the president, only to be reversed in later votes. In Estonia, the discussions over the mode of electing the president (popular vote versus election by the legislature) took up about a quarter of discussion time (Hänni 2002, p. 163) and was debated at 22 meetings (out of 30). In the United States, also, the mode of electing the president was considered to be one of the most troublesome issues, giving rise to lengthy discussions (see, inter alia, Mitchell and Mitchell 1964, p. 108; Crosskey and Jeffrey 1980, p. 23). Again, it was the appeal of Washington that made the constitutional choice of electoral college more acceptable to the Convention (Mitchell and Mitchell 1964, p. 79)—in fact, it was understood that whatever electoral mode would be chosen, Washington would become president—while the prediction that popular elections would lead to the presidency of Rüütel persuaded the Constitutional Assembly in Estonia to opt for electing the president by the legislature.

Testing the importance of partisan interests in establishing electoral rules in the Federal Convention is difficult, since no parties, as conventionally understood, had yet emerged. Also, the framers refrained from establishing in detail how national elections of the legislature were to take place; the electoral procedures were left to be designed by states themselves (Rakove 1997, ch. 8). Still, the group interests were clearly mirrored in the disagreements between the large and small states over how to apportion representation in the legislature and how to elect the president. The conflict over the bases of apportionment in the House and the Senate has been considered by Elster (1995, p. 378, 1998, p. 115) as a prime example of the importance of group interest in constitution-making (see also Elster 1996c; McGuire 1988). Disagreements over how to elect the president also reflected the group interests: the representatives of small states were concerned about their limited influence on the candidate in popular elections and were hence inclined to support the electoral mode where the Senate would be involved or where the vote in the House would be based “one state-one vote” principle (Rakove 1997, p. 259; Mitchell and Mitchell 1964, p. 79; Stewart 2007, p. 155). Framers from large states were for most part more willing to support popular elections, given the advantage it would have granted them in influencing the choice of the president (see Stewart 2007, p. 155).

Testing H4 is again more complicated in the US case than it is in the Estonian case. In the Estonian case, one can clearly distinguish between the external constitutional experts and the members of the CA and discern their constitutional preferences. In the case of the Federal Convention, no “external” experts were contracted or invited to submit their opinions during the summer of 1787. There were, however, some members of the Convention who were viewed as having more “constitutional expertise” than others. The leading constitutional “expert” at the Convention, Madison, argued first for a strong and dominant national legislature, with the other branches—the executive and the judiciary—being appointed by the Congress (Lynch 1999, p. 8). During the course of the Convention’s work, however, partly as a result of the failure of the Virginia plan, Madison started supporting a strong executive alongside the Congress (see Lynch 1999, p. 9) and became a major advocate of the principle of checks and balances. Altogether, one should be cautious in making oversimplified statements like H4; the fact that one is a constitutional expert and well versed in constitutional law does not necessarily mean that one is tilted toward specific constitutional preferences. Importantly, the main problems the constitution-making attempts to solve would play an important role in what the experts recommend. In Madison’s case, the proposals of the Virginia plan were aimed at establishing a stronger central government in a situation where the weakness of the existing legislature had led to a series of problems, threatening the sustainability of the Confederation. Further, the preferences of constitutional experts (analogously to the preferences of other framers) may shift in the course of the drafting process, especially if the decision-making is sequential and formulating a particular constitutional preference is contingent on how other institutional elements have been designed.

While H5 was confirmed in the Estonian case, the US case provides only ambiguous evidence for it. On the one hand, indeed, more than half of the fifty-five delegates in Philadelphia were lawyers (though not constitutional lawyers) and the Supreme Court’s judges are granted extensive guarantees of independence (permanent tenure and irreducible salaries). On the other hand, despite the predominance of lawyers in Philadelphia, the Supreme Court received significantly less attention in the Convention than the design of the legislature and the president. Furthermore, although the discussions in the Convention indicate that a number of prominent framers (like Madison, Wilson, Mason and G. Morris) favoured judicial review during the debates of the Convention, they refrained from including explicit references to that authority in the Constitution itself (Mitchell and Mitchell 1964, p. 230). Also, although a number of lawyers (like Wilson, Ellsworth and G. Morris) argued for giving the Supreme Court the right to join the executive in “putting a negative on acts of the legislature”, other lawyers (especially Rutledge and Strong) were against it (Mitchell and Mitchell 1964, p. 87).

6 Conclusions

Although a number of theoretical discussions on constitution-making have attempted to put forth nomothetical hypotheses on how the kinds of participants involved in constitution-making influence the proposed constitutional provisions pertaining to the powers of different branches, the case studies in this paper indicate that most of these hypotheses—despite their intuitive plausibility—find only partial support. While most of the hypotheses were at least partially corroborated by the Estonian case, they were either refuted or supported only by weak evidence in the US case. There was only one theoretical proposition—predicting that the constitutional preferences concerning electoral rules and modes of representation will be driven by group interests (either territorial or partisan)—that found consistent confirmation in both cases.

The hypothesis that members of constitutional assembly who come from existing legislature will seek to give overriding importance to the legislature—at the expense of the executive and the judiciary—finds only partial support. As both case studies demonstrate, a framer’s previous (or parallel) membership in a legislative body does not necessarily lead to that framer preferring a strong legislature in the new constitution. In the Estonian case, a number of delegates from the existing legislature argued for constraining the powers of parliament and supported establishing a strong institution of president in the new constitution; in fact, the existing legislature as a whole supported a constrained legislature and a stronger president. Furthermore, in the Estonian case, even those framers who in general argued for a strong legislature did not seek to strengthen the legislature at the expense of the judiciary. None of the framers disputed the importance of including strong guarantees of independence and extensive powers of constitutional review for the judiciary in the constitution. In the US case, although the framers who had been members of the Confederation Congress for most part supported strengthening the national legislature (in fact it would have been almost impossible to make it even weaker), their constitutional preferences clearly diverged on how far that should go. Among other factors, their negative experiences with the excesses of state legislatures made many framers sceptical of granting the national legislature unspecified powers and inspired them to include checks and balances between the different branches in the constitution.

The hypothesis that the incumbent president (or a holder of an equivalent office) will seek to promote a strong presidency found confirmation in the Estonian case. In the US case, however, Washington—the most likely first president—refrained from actively promoting strong powers of the executive during the deliberations of the Convention.

The theoretical proposition stating that group interests (either partisan or territorial) would play a major role in constitutional preferences over electoral rules and bases of representation found confirmation in both cases. Interestingly, in both cases, the mode of electing the president was one of the most contentious issues. Also, in both cases, the framers designed the powers of the president and the mode of electing the president while keeping in mind a concrete person, who was considered to be the most likely first holder of that office.Footnote 10 Thus, it appears that when designing an “individualised” institution, framers find it especially difficult to confine themselves to abstract theorising and to refrain from designing an office to fit a specific person they have in mind.

With regard to the constitutional preferences of constitutional experts involved in constitution-making, the case studies show that simple predictions put forth by Voigt (2004)—stating that constitutional experts will favour constraining the powers of the legislature and granting more extensive powers to the executive and judiciary—may not necessarily hold. While the proposals of the local constitutional experts in Estonia corroborated Voigt’s prediction that experts refrain from attributing extensive powers to the legislature and support granting stronger powers to the judiciary and the executive, the foreign experts giving advice in Estonia warned against a too strong president and advocated establishing a strong parliament. In the US case, the leading constitutional expert Madison also favoured strong powers of the national legislature—at least initially. As both cases demonstrate, the main problems of the polity that the constitution-making attempts to solve would play an important role in what the experts recommend.

Attributing specific constitutional preferences to particular kinds of framers can be especially tricky when a constitutional preference with regard to the powers of a particular branch of government is strongly contingent on other constitutional choices, like the mode of representation. The degree to which the framers’ constitutional preferences shift, depends on whether the institutional choices are made sequentially or simultaneously. While in the Estonian case, the adoption of an entire draft constitution as a basis for further deliberations reduced “sequential” elements in constitution-making (the exception being the powers and the election rule of the president), leading to relatively stable constitutional preferences of individual framers, the sequential constitutional decision-making in Philadelphia makes the application of the list of hypotheses linking a particular kind of framer to a particular constitutional preference very problematic.

An important lesson that can be drawn from the case studies in this paper is that the complex mechanisms involved in constitution-making can render it very difficult for an analyst to capture the constitutional preferences of different kinds of framers in the form of such simple hypotheses as put forth in the existing literature. As the case studies indicate, for most part, no straightforward “yes” or “no” statements could be given about whether a hypothesis holds or not. Hence, one should be cautious in using these hypotheses as building blocks for a theory on constitution-making: such conjectures may conceal significantly more than they reveal. Another relevant lesson drawn from these cases is that although the notions of institutional and partisan interest are appealing since they allow one to put forth parsimonious theoretical propositions, confining oneself to these notions when explaining constitutional design choices of particular countries may provide a rather distorted view of constitution-making. Both in Tallinn and in Philadelphia, the constitutional choices of the framers were influenced by constitutional interests and constitutional theories—and complicated interplays between them.