Abstract
Institutional interests are often the main determinant of day-to-day politics. However, do they also matter in the more consensus-oriented field of constitutional politics? To answer this question, this article examines the success and failure of constitutional amendment drafts. We reassess a hypothesis proposed by Donald S. Lutz more than 20 years ago, according to which the initiator of an amendment is a significant determinant of its success, that is, of its passing or not passing. This study is based on a unique dataset of successful and failed constitutional amendments, covering 18 post-socialist countries in Central and Eastern Europe (1990–2014). We demonstrate that the chances of success for a given constitutional amendment are clearly driven by institutional interests: cabinet and presidential proposals have significantly higher chances of success than parliamentary and public initiatives. Additionally, success or failure also depend on the level of democracy and the rigidity of the amendment process.
Keywords
Introduction
All over the world, from Hungary to Egypt, in Mexico or Botswana amending constitutions is a key tool in the transition from autocracy to democracy; unfortunately, also from democracy back to autocracy. In this time of an astounding decline in democratic regimes that amounts to a new wave of autocratization, understanding the way in which constitutions are amended and what makes amendment drafts more, or less, likely to be adopted, is particularly salient.
Obviously, not every amendment draft is approved or implemented. In fact, the rate of successful constitutional amendments, that is, amendment proposals that have been passed and have come into effect, varies dramatically from one country to another, and often in ways that are not noticeable on first glance. For instance, the number of amendments adopted in the Czech Republic and Croatia are quite similar (Fruhstorfer and Moormann-Kimáková, 2016; Seha 2016). Since the first post-socialist constitutions of the two countries came into force, eight amendments have been adopted in the Czech Republic from 1993 to 2014, while six were adopted in Croatia from 1990 to 2014. This could lead one to assume that both countries share a similar level of constitutional activity but in fact, there were 84 attempts to amend the Czech constitution over the above-mentioned period, while there were only eight amendment initiatives in Croatia. This results in dramatically different success rates for the two countries, of 0.10 and 0.75, respectively, which mirrors the variation in the constitutional life of states in the twentieth and twenty-first centuries. But what affects the likelihood that a constitutional amendment will be adopted?
In this study, we argue that political actors have to reconcile two realities while pursuing the amendment of constitutions. On the one hand, they are actors with partisan affiliations, who emphasize their own profile. However, they also function within an institutional logic and expectations of the interests of this specific institution. From a constructivist perspective, these interests vary due to the intrinsic characteristics of the respective institutions, that is, their functions and the way they come into power (Sánchez-Cuenca, 2003: 88). The relationship between these institutional preferences and political outcomes are not unidirectional (as a rational choice argument would suggest). Rather, they interact with the context, thus affecting each other.
This relationship makes it difficult to distinguish between institutional interests and how they translate into institutional behavior and thus political outcomes. This link is neither always strong nor unambiguous. Its strength depends largely on the policy field, the majority constellations, and the timing. Institutional interests ‘constrain and refract politics but they are never the sole "cause" of outcomes’ (Thelen and Steinmo, 1992: 3). This makes it difficult to convincingly relate institutional interests with a political outcome compared to other possible causal factors. But, when a political outcome requires consensus-oriented inter-partisan cooperation beyond the basic majority requirement for normal legislation, we expect institutional interests to become more visible.
This argument engages with an important discussion in research on institutional change. Within the dominant neo-institutionalist framework, constitutions are too often conceptualized as exogenous. Yet, constitutionalism research has provided ample evidence that constitutions are, as other institutions, also endogenous (Pérez-Liñán and Castagnola, 2016). In line with this research, we argue that constitutional change has both endogenous and exogenous causes. We test for both of these and demonstrate that the endogenous, that is, institutional interests, of the initiators are the main explanation for the success or failure of a modification of the (constitutional) rules under which they act.
The field of constitutional politics requires partisan cooperation, since enacting a new constitution or amending it regularly entails qualified parliamentary majorities. Hence, constitutional politics provides an ideal field to link institutional characteristics and institutional interests with political outcomes.
Previous research on constitutional politics has furthermore focused almost exclusively on approved amendments. However, this success-oriented perspective significantly narrows our understanding of constitutional politics. Expanding our view to the largely unexplored realm of failed amendments may provide not only insights into the role of institutional interests but also the key to a more comprehensive understanding of constitutional change. It is difficult to fully grasp the logic of constitutional politics without having any variance in the dependent variable—in other words, without including failed cases of amendment drafts as well. Several authors have raised the question on the success and failure of constitutional amendments and analyzed prominent examples, such as the failed equal rights amendment in the USA (Manfredi and Lusztig, 1998), the rejected federalism reforms in Austria and Canada (Benz, 2016), or a variety of failed amendments in Italy and Norway (Rasch and Congleton, 2006). Nevertheless, knowledge about what leads to the success or failure of constitutional amendments is virtually non-existent, because data for wider comparative studies are difficult to collect. Doing so requires in-depth analyses of parliamentary protocols and related documents in a variety of languages.
To test the relationship between institutional interests and the success or failure of constitutional amendment drafts, we rely on a pioneering—and the only comparative—study on this topic. Donald S. Lutz (1995) examined the development of US state constitutions and analyzed the influence of who submitted a proposal on the amendment’s success. He showed that popular initiatives have lower success rates than amendments stemming from the legislature or from a constitutional convention. Lutz (1995: 254) argued that popular initiatives generally face more hostile reactions from legislatures than the other two types of amendments. Putting this observation into a broader context of differing institutional interests in political systems, we argue that the probability that a constitutional draft amendment is adopted, depends on its initiator, since institutional interests vary among different initiators, involve different submission behaviors, and invite differing treatments of submissions.
Lutz’s analysis compared a set of very similar cases and disregarded any control variables. Thus, his study might suffer from the ‘third factor’ problem or just illustrate a US peculiarity. Therefore, we tested an extended version of Lutz’s hypothesis in a different comparative setting. Our study is based on a new and unique dataset of constitutional amendments in 18 post-socialist countries in Central and Eastern Europe (CEE). The data were collected as part of the research network ‘Constitutional Politics in the Post-Socialist States of Central and Eastern Europe’ (Fruhstorfer and Hein, 2016b). The data cover the period ranging from the enactment of the first post-socialist constitution after 1989 in each country to the end of 2014. 1 It is the first dataset of its kind. Focusing exclusively on CEE countries has two main advantages. First, these cases include various types of governmental systems from parliamentary to premier–presidential, as well as president–parliamentary systems (Elgie, 1999), different levels of democracy, and different levels of constitutional rigidity. This constitutes a sufficient variation of the main control variables. Second, focusing on the transitional societies in CEE allows us to analyze a set of countries where the institutional interests are less mitigated than in other—more historically grown—constitutional systems.
Based on this new and unique dataset, we systematically analyze the effect of individual political institutions on the probability that constitutional amendments are adopted. The results show that the chances of success for a given constitutional amendment are clearly dependent on the institution that proposed the amendment. We argue that the necessity for consensus above (not necessarily instead) partisan considerations in the amendment process allows us to make a clear distinction between institutional and other forms of interests of the involved actors.
The approach developed in this article opens the door to a more accurate understanding of how constitutions are changed, what role the introducing institution plays as an endogenous political element, and why so many amendment proposals are made without any hope of success. Our results narrow the gap in the understanding of constitutional politics that opened up with the sole focus of empirical research on successful amendments. This has important implications for the study of interest formation beyond clear power politics and constitutional development under varying conditions.
Theoretical expectations
Several factors lead to constitutional amendments. Although it is impossible to account for all of them, a general logic underlies many of these initiatives. Designing constitutions or changing them requires more effort than the general legislative process, yet the constitutional initiative process is still driven by actors seeking a bargaining advantage (see Elkins et al., 2009). As a result, majorities reform the constitutional order when they see an opportunity to advance their future position (Rasch and Congleton, 2006: 538f.). Minorities also use constitutional amendments to seek advantages—whether directly, by getting amendments approved, or indirectly, by influencing the public debate and receiving advantages in other areas. In addition, the public might have an interest in constitutional amendments. This can be triggered by changing societal cleavages that are no longer met by the extant constitutional design or by a desire to either constrain or empower government (Rasch and Congleton, 2006: 539). These are all important aspects, but it is equally important to consider that constitutional amendments are pursued by institutional actors. The sheer existence of proposals that have little chance of success shows that success does not necessarily have to be the main motivation. Assumptions about the motives of the actors involved and how they interact with the institutional framework and influence the outcome are therefore crucial. Who, then, starts the amendment process and how consequential are institutional interests and the context on amendments’ success?
Who initiates constitutional amendments—and why?
Institutions have both an exogenous role of mediating between individual preferences and available choices and an endogenous role of ‘molding and channeling preferences’ (Shepsle, 1986). What is at times forgotten is that institutions also have interests of their own. In a general sense, ‘different branches of power, due to their different modes of election and their different functions and capabilities, have different institutional interests’ (Sánchez-Cuenca, 2003: 88). Hence, the extent to which individual preferences turn into political decisions is contingent upon the interests of the institutions involved. We think that the key actors in constitutional amendments, that is, parliamentary deputies, cabinets, and presidents, have interests that not only derive from their partisan affiliations or their ideological convictions, and are not automatically based on power asymmetries. As we know, these asymmetries are not even a clear and undisputed explanation in understanding ordinary politics or legislation (Döring, 2017). It thus applies even less in the area of constitutional politics, where bipartisan activities are typically needed in order to reach qualified parliamentary majorities. Jon Elster (1995) confirmed this observation for the field of constitutional politics and design. He showed that institutional interests drive the development of individual institutional configurations. However, research also tells us that, in particular, ideologically homogenous interests of parties trump institutional interests in pursuing political outcomes (Balkin and Levinson, 2006: 520). To understand the success and failure of constitutional amendment drafts, we must consider these two perspectives on the impact of institutional interests.
Lutz (1995) showed that the success of constitutional amendments is influenced by the interest of the pursuing institution. In his study on the development of US state constitutions, he finds that popular initiatives have a lower chance of success than draft amendments originating from either the legislature or a constitutional convention (Lutz 1995: 254).
When transferring Lutz’ (1995) hypothesis to CEE, we must consider a different set of possible initiators, namely parliaments, presidents, cabinets, and citizens. 2 Regarding this set of actors, the following assumptions seem plausible: in parliamentary and semi-presidential systems of government, the cabinet usually controls its own majority in parliament. Therefore, it is in a good position to negotiate with the opposition to achieve a qualified parliamentary majority, which is needed to adopt a constitutional amendment in 16 of our 18 cases. 3 In addition, governing parties are not interested in losing a vote, since this could be publicly interpreted as a victory for the opposition. We therefore expect cabinets to submit an initiative only if they presume that they have a real chance of passing that amendment in parliament. Their institutional interest is thus to avoid humiliation and the undermining of their position of power. The same should apply for presidents—at least if they belong to the governing majority. Even in the case of “cohabitation” (Elgie, 2010: 29), a president will probably not be interested in suffering a public defeat and would therefore use the right of initiative very carefully.
In contrast, we assume that parliamentary (i.e., usually oppositional) and popular initiatives will have comparatively low success rates. These actors are not necessarily interested in (if even capable of) securing a parliamentary constitutional majority before submitting their amendment. As with ordinary legislation, amending the constitution is not the only goal parliamentary actors pursue. For instance, ‘backbenchers within the governing party or parties’ camp may use legislative activity to raise their profile and garner support from their constituents to enhance their chances of re-election or to capture more influence within their political party’ (Bräuninger and Debus, 2009: 805, 833).
Thus, the institutional interests of parliamentary deputies or groups are attention, control, and information—in other words: the basic functions of a parliament. Parliamentary behavior and agenda setting in parliamentary and semi-presidential systems are often characterized by a clear separation between the government and the parliamentary majority backing it on the one side and the opposition on the other side. At the same time, the separation of powers has two dimensions: functionalist; and institutionalist. Given the context of constitutional change, we argue that the logic of institutional separation prevails: parliamentary initiatives for constitutional amendments mainly come from deputies, who propose them to discuss certain topics in the public or parliamentary arenas, or even to shame the majority for being inactive. They follow their institutional interests of informing the public and raising attention to certain topics. In doing so, deputies do not abide by their partisan affiliation but are also influenced by legislative rules (Cox, 2000; Zubek, 2008), inter-party dynamics (Müller and Strøm, 2003), and their attitudes (Szczerbiak, 2008).
Furthermore, we have to consider popular initiatives. They usually address topics and ideas that are not present or are at least marginalized in the parliamentary arena. Citizens might even act against a large majority of the political elite to criticize them and expose them as distanced from the people’s views and preferences. Consequently, we expect popular initiatives to have comparatively low chances of success, especially if there is a need for parliamentary approval.
In sum, our key hypothesis extending Lutz’s (1995) considerations is this: The probability that a constitutional draft amendment is successfully adopted is higher when it is initiated by the cabinet or the president than by parliament or by the public.
The subgroup of parliamentary initiatives requires some further consideration. We expect parliamentary initiatives for constitutional amendments to stem mainly from opposition parties. Their opportunity structures are mainly shaped by the quorum for proposing an amendment, which we formalize as a variable we call parliamentary initiative rigidity. In the constitutional orders under study, these quorums differ greatly: from a single member of parliament (e.g., in Bosnia and Herzegovina, and the Czech Republic), to one-third of all deputies (in Montenegro, Serbia since 2007, and Ukraine). A low parliamentary initiative rigidity makes the typical opposition usage of parliamentary motions easier. Therefore, we expect it to increase the number of amendment initiatives. However, the chances of success of such draft amendments might be rather low because they will be opposed by the institutional interest of the governmental majority to reject any oppositional proposal independent from its content. In contrast, a high level of parliamentary initiative rigidity makes the outlined oppositional kind of behavior relatively unlikely and increases the chance that only amendments with a realistic chance of success will be put on the parliamentary agenda.
Thus, our additional hypothesis regarding parliamentary initiatives is: The probability that an amendment initiated by parliament is successfully adopted increases with the parliamentary initiative rigidity, that is, the share of members of parliament necessary to propose an amendment.
Control variables
To test these hypotheses, we control for other potential contributing factors. First, we incorporate the content of the amendment along the lines of key political interests. It is impossible to build a variable that fully represents the complexity of constitutional amendments’ contents. Hence, we differentiate between amendments that would change the balance of power between state bodies, and amendments modifying (in most cases: enhancing) fundamental rights and freedoms. The former directly affect the political power of the participating groups and institutions. In most cases, this equals a zero-sum situation, where any change would increase the power of some participant and decrease the power of another one, thus making a more fine-grained coding scheme not very useful (a claim in which we follow other authors, for example Dixon, 2018). In contrast, the latter kind of amendments does not (at least not directly) affect the balance of power between state bodies and political groups. In addition, enhancing fundamental rights and freedoms often meets with approval by societal majorities. In sum, it seems reasonable to expect that amendments changing the balance of power between state bodies and political groups are harder to get approved than amendments modifying fundamental rights and freedoms. As our main hypothesis for the success or failure of amendment proposals focuses on the type of initiator, we also consider interaction effects between content and initiator.
Next, we control for those variables that create the context in which the constitutional amendment occurs as we aim to distinguish the contextual factors from the institutional interests. The most common, but also most controversial, explanatory variable for constitutional political outcomes is the formal rigidity of the constitution (Lorenz, 2005). It is reasonable to assume that a higher rigidity of the constitution, that is, a higher difficulty to amend the text, leads to a lower amendment frequency. However, comparative empirical studies have found contradictory results for different regional settings. Relevant studies for post-socialist CEE show that formal amendment rigidity does not have a significant impact on the frequency of adopted amendments in that region (Lorenz, 2011; Roberts, 2009).
However, when it comes to the success of constitutional amendments, the picture might look different. According to Christopher P. Manfredi and Michael Lusztig, a high level of rigidity makes it hard for political actors to reassess amendments later and address mistakes in the case of, for example, unintended consequences or unexpected interpretations during constitutional review. Therefore, actors will tend ‘to make their demands in unambiguous language that is invulnerable to interpretations that may run counter to the sponsoring group’s intent’ (Manfredi and Lusztig, 1998: 382). This, in turn, makes compromise between political actors more difficult, and ‘[t]he result is constitutional failure through amending process overload’ (Manfredi and Lusztig, 1998: 399). This assumption implies an indirect self-fulfilling prophecy, whereby the more rigid the constitution is, the less likely it will be that amendments will be approved.
In addition, rather rigid constitutions make it easy to blame other actors for obstructing reforms. Under such circumstances, amendments can be proposed to force a public debate on certain topics or even to expose competing political actors, not despite but because these initiatives are seen as having no chance of succeeding. In essence, it seems reasonable to assume that the more rigid a constitution is, the less likely it will be that amendment drafts will be approved.
Another variable that might influence the success or failure of constitutional amendment drafts is the level of democracy. So far, research on constitutional politics has almost exclusively focused on either democratic or non-democratic political systems but only rarely on both at once (Ginsburg and Melton, 2015). However, the difference, between democratic and autocratic regimes (or the position of a given regime on the continuum between those two ideal types) might affect the chances of success of constitutional amendments. Autocratic presidents can be expected to be able to get any amendment that they deem necessary approved. In addition, it is rather unlikely that there would exist an opposition strong enough to initiate constitutional amendments even if they were formally allowed to do so (see Linz, 2000: 159ff). In contrast, democracies show both a much higher degree of uncertainty about the outcome of political decisions and a much higher level of pluralism regarding the interests of political actors. Thus, the probability that a constitutional amendment is successfully adopted might decrease with the level of democracy of the political regime.
Next, as Andrew Roberts (2009: 101) and Astrid Lorenz (2011: 62) have demonstrated, the political fragmentation of the parliament negatively affects the frequency of adopted constitutional amendments. We expect a similar influence on the chances of success of amendment initiatives since higher fragmentation might make it more difficult to gain the necessary qualified majority. This is particularly plausible in parliamentary and semi-presidential systems, where bipartisan activities bringing governmental and oppositional parties together is not a typical feature. In sum, the probability that a constitutional amendment is successfully adopted might decrease with the political fragmentation of the parliament.
We also include the level of participatory constitution-making as a control variable. We base this on an assumption made by Lorenz (2011: 61): if a constitution was agreed upon by a special body and/or put to a referendum, one should expect the first constitutional amendment to be approved later than in cases of simple parliamentary constitution-making, since more (formal) legitimacy results in more commitment toward a public voice in constitution-making. We transfer this assumption to the success of constitutional amendments: it might be more difficult to amend a constitution that changes the (implicit) will of a more inclusive forum (i.e., the people) than a constitution that only draws on the same degree of public approval as the amending institution itself.
Roberts (2009: 101) has also argued that the age of a constitution might exert an influence on the frequency of amendments. As the original context of the constitution-making changes, ‘the constitution will lose touch with current circumstances. Political actors will in turn try to adopt amendments that allow them to better deal with the problems of the day’ (Roberts, 2009: 101). This relation might likewise affect the chances of success of amendment proposals, that is, the older a constitution is, the higher the chances are of an amendment proposal being adopted.
Our cases include various types of governmental systems from parliamentary to premier–presidential, as well as president–parliamentary systems. Since the type of the governmental system supposedly influences the probability of an amendment draft being successfully adopted, we include this as a control variable, as well.
Finally, we include accession to the European Union (EU) as a control variable, considering the specific context of post-socialist CEE. Eleven of the 18 examined countries entered the EU during the period under study. For that reason, many of these countries had to revise their constitutions in several regards, most notably by introducing EU clauses reforming the judiciary, improving human rights protection, and restricting office holders’ immunity rights (Fruhstorfer and Hein, 2016a). Since becoming an EU member state was the consensual goal of a vast majority of the political elites in those states, the chances of success of constitutional amendments might have been higher during the pre-accession period.
Data
We draw on an original dataset that comprises 18 post-socialist countries in CEE, covering the period from the implementation of their first post-socialist constitution until the end of 2014. The dataset is based on detailed analyses of parliamentary protocols, committee documentation, constitutional court rulings, and other related archival material. The dataset includes all 557 detectable successful and failed constitutional amendments in these countries. Each draft amendment was coded as the year of its introduction.
The unit of analysis is amendment laws. The group of successful amendments comprises all adopted amendments. The group of unsuccessful amendments consists of all proposed amendment laws that reached at least the first step of the formal amendment procedure as outlined by the valid constitution but failed later on in the process—either in a parliamentary vote, by exclusion from the parliamentary agenda, by expiration due to the end of the legislative period, by not reaching the necessary quorum or majority in a referendum, or by being declared unconstitutional by the constitutional court (in cases where the court was involved ex officio in the amendment procedure as a regular veto player according to the constitutional amendment formula). We did not include proposals that were withdrawn by their initiators or proposals to enact a new constitution. The success (1) or failure (0) of every individual constitutional amendment initiative was coded as the dichotomous dependent variable ‘success’.
The independent variable initiator is coded as a dummy variable for four different institutions: the parliament; cabinet; president; and people. Joint initiatives by two actors were coded twice—once for each initiator, that is, each equaled one data point in the dataset. 4 The independent variable parliamentary initiative rigidity is constructed as a categorical variable from 0 to 1. In case every single deputy can propose an amendment, we assign 0 points. If, conversely, a unanimous vote of all members of parliament is necessary to initiate an amendment, we score it as 1. The variance in between is captured by the share of deputies required to initiate an amendment. If, for instance, the quorum equals one-quarter of all deputies, we score it 0.25.
The control variable content of the amendment is coded as a dummy variable for amendments affecting the balance of power between state bodies and political groups (i.e., regulating the state organization, the competencies of state bodies, and the rights of office holders), and amendments modifying fundamental and civil rights and obligations (including political rights of the citizens, such as voting rights).
The rigidity of the constitution is measured by means of the index provided by Lorenz (2005). Among the cases studied, the values vary from 3.0 (several countries, easy to amend) to 10.0 (Romania since 2004 and Russia, difficult to amend). Admittedly, one-dimensional rigidity indices fail to track the empirical reality in cases of ‘tiered constitutional design’, that is, constitutions with ‘different rules of constitutional amendment for different parts of the constitution’ (Dixon and Landau, 2018: 438). Although this applies to a majority of constitutions in our dataset, there has been only one failed (in Lithuania 2014) and one successful amendment (in Estonia 2003), to which one of the more demanding amendment procedures codified in those countries’ constitutions applied. Hence, the use of a one-dimensional rigidity index—all its deficiencies notwithstanding—is reasonable for our empirical data.
The level of democracy is measured with the Freedom House index (FHI; Freedom House, 2016), a well-established index used to measure the empirical level of democracy. It ranges from 1.0 (free) to 7.0 (not free), with 6 points for Belarus and Russia in various years as the highest value in our dataset (indicating the least-free countries). For political fragmentation of the parliament, we applied the effective number of parliamentary parties for the year in which the amendment was proposed, using data provided by Michael Gallagher (2015). The level of participatory constitution-making is coded dichotomously with it equaling 1 if the constitution was agreed upon by a special body and/or put to a referendum and equaling 0 if written by a parliamentary body. We also controlled for the type of the governmental system (parliamentary and semi-presidential systems) (Elgie 1999). We also code the accession to the EU, with the pre-accession period (five years prior to EU accession as well as the year of accession). The age of a constitution is measured in calendar years after its enactment.
Results
Successful and unsuccessful constitutional amendments: An overview
In total, the 23 constitutions in our 18 CEE countries had been the targets of 557 amendment initiatives by the end of 2014. These initiatives resulted in 128 amendments (success rate = 0.23). Regarding both the number of initiatives and those successful amendments, there is considerable variation between the polities under study (see Figure 1). While Hungary has seen 118 constitutional amendment proposals and 34 successful amendments from 1990 until 2014, no amendment initiative has been recorded in the same period in Serbia. 5 The success rates vary even more, ranging from 0.10 in the Czech Republic to 1.00 in Montenegro and Russia.
Successful and failed constitutional amendments in Central and Eastern Europe per country (1990–2014; deviations due to joint initiatives).
Out of these 557 initiatives, parliaments had put forward 442 proposals, while cabinets initiated 85, presidents 23, and citizens seven (see Table A1 in the Online Appendix). As expected, this is largely driven by the parliamentary initiative rigidity: the lower the share of parliamentarians necessary to submit a draft amendment, the higher is the number of amendment proposals per year (Pearson’s r = −0.55, p < 0.01). The type of initiator and the amendment success are related as expected: parliamentary (success rate = 0.17) and popular initiatives (success rate = 0.14) performed relatively poorly, whereas initiatives from cabinets (success rate = 0.52) and presidents (success rate = 0.48) were by far the more successful on average. These rates are lower than one would expect given the incentives and powers described for these institutions.
The content of the amendment varies considerably. We counted 357 amendment proposals concerning state organization and 90 concerning fundamental rights and freedoms, but there is no variance in the distribution among the initiators. In both fields, around 80% were initiated by the parliament.
Explaining the success of constitutional amendments
To ascertain the effect of the initiator and thus the institutional interests on the likelihood of a proposed constitutional amendment being successful, we apply a logistic regression analysis. All correlations between any explanatory and control variables are well below the level at which multicollinearity would become problematic. We also tested for specification errors to avoid omitted variables. We begin with a conventional logit analysis for the whole dataset in Model 1 (see Table 1). Because our variables of interest are measured at the country level while the dependent variable is measured at the amendment level, we clustered the standard errors at the country level. The model correctly predicts 79.2% of the outcomes and performs well in terms of goodness of fit with a C-statistic of 0.70.
Analysis of the success of constitutional amendments.
Notes: Logistic regression. Robust standard errors in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01.
As the estimates of Model 1 and Model 2 show, the type of initiator plays an important role. As expected, the odds of amendments being successful vary dramatically and significantly depending on the initiators. The differences confirmed our assumption that draft amendments are more likely to be adopted when they are initiated by the cabinet or the president than by the parliament or public. All else being equal, the chances of a cabinet proposal being adopted are 6.78 times higher than those of a parliamentary initiative. Likewise, presidential initiatives have a 4.16 times higher chance of succeeding compared to parliamentary initiatives. In contrast, the odds of public initiatives do not significantly vary from parliamentary proposals. These findings support our assumption that parliamentary initiatives mainly come from opposition parties, whose primary institutional interest is not the actual change of the constitution but to place certain topics on the public or parliamentary agenda, or even to shame the majority for being inactive. Thus, the different institutional interests condition the likelihood of an amendment proposal being successful.
From the eight control variables introduced in our analysis, two variables are significant and show the expected direction of the influence. First, the formal rigidity of the constitution negatively affects the chances of getting an amendment proposal adopted. All else being equal, the odds of an amendment being successful decrease by 20% for every point increase in the Lorenz’ index of amendment rigidity. Second, the success of amendments is also largely driven by the level of democracy. Every point increase on the FHI (which translates into a lower level of democratic freedom) increases the odds for a successful constitutional amendment by 59% (likewise ceteris paribus). This is no surprise given the fact that constitutional amendments in the non-democratic countries under study, particularly Russia and Belarus, are not seriously up for discussion. In contrast, neither the political fragmentation of parliament, the level of participatory constitution-making, the age of the constitution, the type of the governmental system, nor the accession to the EU affects the success or failure of constitutional amendment proposals. Likewise, the content has no significant effect on the probability that an amendment is successful. Since we have been able to code the content of close to 80% of all amendment drafts in our dataset due to restricted data availability, we excluded that variable in Model 2 in order to analyze all 552 6 cases (instead of 442). This slightly changes the results regarding other control variables: Whereas the variable rigidity loses its statistical significance, the political fragmentation of the parliament appears as a significant regressor in the theoretically expected direction. Model 1 and Model 3 were also tested with an interaction effect relating the initiator hypothesis to the content of the amendment proposal (see Table A2 in the Online Appendix). The interaction effect is not significant and indicates no systematic interaction between the content of the amendment and the initiator when it comes to the success of these amendments. 7
Only 58 out of the total of 557 observations come from hybrid or autocratic regimes. However, we have assumed that the effect of the independent variables initiator and rigidity might be substantially restricted in autocratic and hybrid regimes, leading to a generally higher share of successfully adopted constitutional amendments in these countries. 8 This is the case, indeed. Whereas in democracies (free according to the FHI), only 106 out of 499 amendment proposals were adopted (success rate = 0.21), non-democracies (partly free and not free according to the FHI) show a considerably higher success rate (0.38), with 22 out of 58 amendment initiatives adopted. In those countries, presidential initiatives are particularly successful: 10 out of the 16 amendments proposed by the heads of state of Belarus, Croatia (non-democratic until 2000), Moldova, Russia, and Ukraine (non-democratic except from 2005 to 2009) were adopted (success rate = 0.63). More concretely, Presidents Alexander Lukashenko (Belarus, in power since 1994), Franjo Tuđman (Croatia, 1990–1999), Dmitry Medvedev (Russia, 2008–2012), and Vladimir Putin (Russia, 1999–2008 and since 2012) did not fail to pass a single amendment proposal either in parliament or via referendum. These leaders were clearly able to use the constitution as an instrument of power in polities that are not characterized by the rule of law but rather by the rule by law.
To address the obvious sensitivity of the results to the level of democracy, we excluded the observations for all non-democratic countries, that is, those countries that received more than 2.5 points on the FHI scale in the year of the constitutional amendment in question. Models 3 and 4 in Table 1 thus present the results for the subgroup of 499 amendment proposals in the democratic countries under study. The model correctly predicts 82.7% of the outcomes. The effect of the independent variable initiator is again of considerable magnitude in the theoretically expected direction, the odds even being slightly higher, as in Models 1 and 2. Ceteris paribus, the chances of a cabinet proposal being adopted in democracies are 7.83 times higher than those of a parliamentary initiative (Model 3). Presidential initiatives even have an 8.50 higher chance of succeeding compared to parliamentary initiatives, whereas the odds of public initiatives in democracies do not significantly vary from parliamentary proposals.
From our set of control variables, the rigidity of the constitution appears again with a significant effect on the chances of success of amendment proposals in line with our theoretical expectations. All else being equal, the odds of an amendment being successful decreases in democracies by 42% for every point increase in Lorenz’ (2005) rigidity index (Model 3). In contrast, none of the other control variables significantly influence the chances of success of constitutional amendments in democracies. The political fragmentation of the parliament appears again as a significant regressor when we do not control for content (Model 4), and it shows no difference between democracies and non-democratic systems. Similarly, the differences in democratic quality within this subset of “free” countries have no significant impact. Somewhat surprisingly, the accession process to the EU also does not have any influence on the success or failure of constitutional amendment drafts, although that process has shaped the constitutional orders of the respective countries to a considerable extent.
In order to test our additional hypothesis on the parliamentary initiative rigidity, we present the results for the subset of parliamentary initiatives in Models 5 and 6. We argued earlier that parliamentary amendments stem mainly from the opposition. They are not primarily aimed at changing the system but at putting forward a different agenda and gaining attention. This effect is particularly observable in those democracies where each member of parliament can initiate a constitutional amendment, that is, in Hungary, the Czech Republic, Slovakia, and Poland. Such a pattern is likewise noticeable in Latvia, where only five deputies are needed for an amendment initiative.
However, Model 5 points to the missing statistical significance of the parliamentary initiative rigidity. Instead, the general constitutional rigidity is a stronger explanation for the probability of a successful parliamentary draft than the number of deputies necessary to propose it. This means that the parliamentary initiative rigidity decreases the probability of parliamentary amendment proposals (as shown above), but it has no influence on their chances of success. If a constitution requires a high parliamentary quorum for submitting a constitutional amendment draft (e.g., one-third of all deputies), this will lead to a comparatively low number of proposals, but the chances of each proposal being adopted is not higher than in cases where a constitution entitles every single deputy to submit an amendment draft. Of the control variables, the mode of constitution-making reaches significance.
In addition, both the level of democracy and the level of participatory constitution-making appear as significant regressors in Model 5, in both cases in the theoretically expected direction. When we exclude the variable content in order to include more cases into the analysis (Model 6), both of these control variables as well as rigidity lose their statistical significance. In contrast, the accession to the EU becomes significant.
Conclusion
This study has shown that the institution that initiates a constitutional amendment is the most important determinant of its success or failure. Whereas proposals initiated by the cabinet and the president have comparatively high chances of success, parliamentary and public initiatives are likely to fail. The analysis also revealed that the rigidity of the amendment procedure has a significant negative effect on the probability that a constitutional amendment will be successfully adopted. In other words, the most-debated factor in the research on amendment frequency appears as a significant explanatory variable regarding the success or failure of amendment initiatives. In contrast, the content of these amendment initiatives does not influence their probability of being successfully adopted. To our surprise, we also found no statistical evidence that a parliamentary draft’s chances of success increases with parliamentary initiative rigidity, that is, the share of members of the legislature necessary to propose such a draft.
This has major implications for our understanding of how the fundamental rules of political systems are changed. Most importantly, amending constitutions is hardly ever a pure partisan and majority-backed decision. The process usually requires more cooperation of different political actors than ordinary legislation does. As a matter of course, a statistical analysis is not able to identify the causal mechanisms behind this pattern. However, what we can discern about the differing institutional interests and political strategies of the four types of initiators under study make this pattern plausible.
This finding has important implications for theories of institutional change and constitutional politics. The few studies that have engaged with the idea of specific institutional interests in constitutional design and change have almost exclusively entertained the notion that institutional interests are self-serving interests. But several empirical examples provide a different perspective on the role institutions play in the process of constitutional design beyond simple self-interest. If one follows our argument that the more-consensus requirement of constitutional amendments makes institutional interests more visible, the findings about the role of different initiators for the probability of amendment success have important implications for our understanding of institutional interests.
There is much more to be learned about the relationship between endogenous and exogenous institutional interests, partisan affiliation, personal motivations, and the success of constitutional amendments. Consequently, future conceptual efforts and empirical analyses should extend our work. Specifically, these efforts should focus on the question of whether our findings are generalizable to other world regions, governmental systems, and research periods.
The ideas developed, and results presented in this article have paved the way to a more accurate understanding of constitutional politics that systematically takes into account institutional interests and the formal conditions of amendment processes. In doing so, the article has narrowed the gap in our knowledge about constitutional amendments and their movement through the amendment process. In particular, it has shed light on the endogenous role of institutions changing the rule of the political game they are playing themselves.
Supplemental Material
sj-pdf-1-ips-10.1177_0192512119871323 – Supplemental material for Institutional interests and the politics of constitutional amendment
Supplemental material, sj-pdf-1-ips-10.1177_0192512119871323 for Institutional interests and the politics of constitutional amendment by Anna Fruhstorfer and Michael Hein in International Political Science Review
Supplemental Material
sj-pdf-2-ips-10.1177_0192512119871323 – Supplemental material for Institutional interests and the politics of constitutional amendment
Supplemental material, sj-pdf-2-ips-10.1177_0192512119871323 for Institutional interests and the politics of constitutional amendment by Anna Fruhstorfer and Michael Hein in International Political Science Review
Footnotes
Acknowledgements
The authors thank all members of the research network ‘Constitutional Politics in the Post-Socialist States of Central and Eastern Europe’ for their participation in the project and collecting the data used in this article. For comments and helpful suggestions, the authors are also grateful to Diana Burlacu, Zachary Elkins, Alexander Hudson, Ran Hirschl, Eugénie Mérieau, and Kelly Rader and the anonymous reviewers.
Funding
This article was written in the context of the Alexander von Humboldt Professorship in Comparative Constitutionalism at the University of Göttingen. The work of the research network ‘Constitutional Politics in the Post-Socialist States of Central and Eastern Europe’ was supported through funds from the Excellence Initiative of the German Federal and State Governments (run by the German Research Foundation together with the German Science Council).
Supplemental material
Supplemental material for this article is available online.
Notes
Author biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
