Political conflict, political polarization, and constitutional compliance

Abstract

While the economic approach to constitutions highlights their contribution to resolving conflict, recent work on the de jure–de facto distinction in relation to various constitutional rules suggests that political conflict and polarization could play a role in explaining the size and evolution of the gap between constitution text and constitutional practice. In this paper, we are interested in the relationship between the degree of conflict in the political arena within the state, captured by the polarization of the political landscape, as well as the underlying political polarization in society, and compliance of government actors with the country’s constitution. Based on a number of theoretical arguments, we provide an empirical investigation for ca. 170 countries in the period 1975–2020, using the new Comparative Constitutional Compliance Database. Our results suggest that constitutional non-compliance is associated with more intense political polarization in society, but it does not seem to be correlated with polarization of the political landscape.

1 Introduction

Constitutional law is not one of the traditional branches of law that has been studied from the economic perspective for decades, as is the case for property, contracts, tort, criminal law, or antitrust. Economic studies related to constitutions have mainly been developing in close connection with Buchanan’s constitutional economics (Buchanan & Tullock, 1962)—a research program viewed by many as part of public choice. Positive constitutional economics concentrates on two crucial questions—regarding the determinants and the effects of constitutions (Voigt, 2011). Only a decade or so ago, scholars active in this field extended their focus to other constitutional issues such as the endurance of national constitutions (Elkins et al., 2009). Most recently the problem of constitutional compliance, in particular the causes and effects of constitutional practice diverging from the constitution text, have become an important new topic of interest.

Current times have brought about numerous examples of situations when constitutional practice of political actors diverged from the formal wording of constitutions. Voigt (2021) surveys the scant literature concerning the causes of the existence of the so-called de jure–de facto gaps with regard to constitutions and proposes a general conceptual framework for their analysis. Rooted in economics, this approach focuses on the incentives that constraints stemming from constitution texts, as well as from the environment within which constitutions operate, generate for government actors. Although Voigt’s (2021) paper offers many conjectures and hypotheses, not a single one directly relates to the role of political conflict nor the underlying political polarization. This is a lacuna that we aim to fill in this paper.

The connection between political conflict and constitutions, in general, has not gone unnoticed in economic analysis. In fact, one of the foundations of the economic approach to constitutions highlights the role of constitutions in coordination and resolving political conflict (Hardin, 2013; North et al., 2009; Weingast, 1997). Indeed, a central function of formal rules, including constitutions, is coordinating collective action around a focal solution, which is an equilibrium in the game between the sovereign and citizens (Hardin, 2013; Weingast, 1997). In line with this view, it is commonly argued that a written text of the constitution can minimize conflict over basic institutions (and this for any regime—both democracy and dictatorship—see Ginsburg & Simpser, 2013). In this paper, we aim to complement this literature with a slightly different focus. Namely, we investigate the role that the degree of conflict in the political arena within the state, captured by polarization of the political landscape, as well as the underlying political polarization in society, can have for the divergence of constitutional practice of political actors from the formal wording of constitutions. To the best of our knowledge, these issues have not yet been subjected to closer empirical scrutiny. To be sure, this is not to say that political conflict has not been recognized as an important determinant of constitutional compliance. The focus until now, however, was mostly on armed political conflict.¹ Thus, the issue of the degree of conflict in the political arena connected with the underlying polarization of the society and the political landscape, has been given much less attention.

A more generalized view of the role of political conflict, perceived as disagreement between political actors within a state (not necessarily involving any armed operations), as well as the underlying political polarization of the society, for constitutional compliance can be drawn from a well-established strand of political science literature concerning the so-called crises of constitutional fidelity (Whittington, 2002). This literature argues that constitutional fidelity crises, defined as situations when political actors threaten to become no longer willing to act in compliance with existing formal constitutional rules or systematically violate these rules, are more likely to emerge at times of intensified conflict in the political arena, as well as in highly politically polarized societies. Establishing a link to recent literature on constitutional compliance (Voigt, 2021), we argue that this is because increasing disagreements in the political landscape and within the society may raise political stakes (Ward & Tavits, 2019) and, in effect, increase the benefits from constitution-violating actions by political actors. At the same time it may also adversely affect citizens’ faith in the necessity to adhere to the constraints provided by the constitution. We aim to test these conjectures empirically.

Our empirical approach involves conducting a panel data regression-based study for the global sample seeking to identify the relationships between conflict in the political arena, on the one hand, captured by the polarization of the political landscape, as well as political polarization of the society, and the degree of constitutional compliance on the other. The time period covered is 1975–2020 and constitutional compliance is measured using the new Comparative Constitutional Compliance Dataset (Gutmann et al., 2024b). This database matches information on de jure constitutional rules with data on their de facto implementation in four areas—property rights and the rule of law, political rights, civil rights, and basic human rights. The results of our empirical study provide a first confirmation of the existence of a link between political conflict/political polarization and constitutional compliance, however only with regard to political polarization of the society. The conclusions of our study may offer additional insights on the recent constitutional crises in several countries, in particular where political polarization of the society is high. That said, it needs to be emphasized that, due to the nature of the data, our results should be read as conditional correlations, and thus should not be interpreted as reflecting a causal relationship.

The paper is structured as follows. In Sect. 2, we provide a conceptualization of constitutional (non-)compliance and its determinants from an economic perspective, as well as present a theoretical and literature background for the analysis of the interlinkages between political conflict, political polarization, and constitutional compliance. Based on these considerations, we formulate a hypothesis, which we test empirically in Sect. 3. This section introduces the empirical model, the data sources, and applied econometric techniques, as well as presents and discusses the obtained estimation results. The paper finishes with conclusions.

2 Conceptualization, theoretical background, and literature overview

2.1 An economic approach to constitutional compliance

Voigt (2021) proposed the first systematic approach to economic analysis of constitutional compliance. In particular, his focus is on the so-called de jure–de facto gaps, which are defined as “non-congruence between provisions explicitly written down in the document called the constitution and the behavior of the top representatives of the various government branches such as cabinet members, legislators, and members of the country’s top court(s)” (Voigt, 2021, p. 1780). As will be clear shortly, such an understanding of constitutional compliance is also well fit for analysis of the relevance of political conflict and political polarization in our context.

Viewed from an economic perspective, the behavior of government actors is driven by their incentives, given certain constraints. The latter may be of two types. The first group are design factors relating to the content or structure of the constitution, substantive or procedural in nature. The second group are environmental factors encompassing the conditions, in which a given constitution operates, e.g., constitutional history of a country, values and norms shared in a society, trust between its members, as well as geographical and climate conditions. Given these constraints, the choices of the crucial government actors may be analyzed in a simple cost–benefit framework.

The benefits from non-complying with the constitution depend, in principle, on three factors (Voigt, 2021). Firstly, what matters is the degree to which government preferences are aligned with the constitution text. The more these preferences diverge from the constitution text, the higher the benefits from non-compliance for the government actors. Similarly, the higher the degree to which the constitution constrains these actors, the more advantages they can have from circumventing the existing rules. Finally, personal characteristics of these crucial actors may also play a role, as different types may draw different benefits from non-compliance. Therefore, mechanisms governing the selection of government members (including self-selection) may also be relevant.

Regarding the costs, in turn, constitutions often stipulate sanctions for non-compliance by government actors (e.g. impeachment procedures). Commonly, they also establish constitutional review in instances where non-compliance takes the form of adopting legal acts which are not in congruence with the constitution. The costs for government actors resulting from non-compliance may also depend on the following three factors (Voigt, 2021). Firstly, the more veto-players there are in a given constitutional system, the higher they will be, as these actors will mutually constrain each other against overstepping their competences. Secondly, costs may also be imposed on the non-complying government actors, usually indirectly, by foreign actors (e.g. consisting in a loss of reputation in the international political arena, as well as, in consequence, a possible loss of domestic reputation). This effect is most pronounced if a given country is a party to numerous international agreements and, in particular, if it is a member of an international organization which explicitly monitors and sanctions constitutional violations (e.g. the Council of Europe with its European Court of Human Rights with regard to human rights violations, or the European Union with regard to the rule of law or democracy). Thirdly, non-state actors may also generate costs affecting the incentives of governments to act in (non-)compliance with the constitution. These may, for example, be citizens at large voting non-complying governments (or individuals) out of office, as well as actions of various nongovernmental organizations.

Constitutional violations are, however, not automatically sanctioned. The probability of being sanctioned for non-compliance depends, inter alia, on the availability of information concerning the content of the constitution (that may depend on the degree of media freedom), the degree of precision of constitutional provisions, the degree of consent with regard to their interpretation, as well as the size/scope of the government (smaller governments are easier to monitor).

As presented above, Voigt’s (2021) conceptualization suggests a number of important conjectures and hypotheses about potential factors imposing effective constraints on government actors in the context of possible constitutional violations. However, only a handful of them have been tested in empirical studies and that usually only for a particular selection of constitutional rules (e.g., bills of rights, or judicial independence).² Furthermore, and most importantly given our focus, we believe that one issue, namely political divisions, is largely missing from this portrayal. Below, we argue that having a closer look at this thread may nicely complement the framework just discussed. This is because the degree of political conflict and political polarization can significantly affect the cost–benefit analysis presented earlier.

2.2 The role of political conflict and political polarization

Why could political conflict or political polarization be relevant for constitutional compliance? To answer this question, we draw on political science literature interested in studying constitutional crises. In an important paper in this area, Whittington (2002) proposes distinguishing between two types of constitutional crises: crises of constitutional operation and crises of constitutional fidelity. The former emerge when political disputes cannot be resolved within the existing constitutional framework, i.e., when following all correct constitutional procedures does not lead to a single determinate outcome (formal operational crises) or when the government is incapable of making political decisions or taking effective political actions that are regarded as necessary at a given moment (practical operational crises). The second type of crises—fidelity crises—emerges, in turn, “when important political actors threaten to become no longer willing to abide by existing constitutional arrangements or systematically contradict constitutional proscriptions” (Whittington, 2002, p. 2109). In such a case, successful maintenance of political disputes between various domestic actors within the bounds set by the constitution is dependent on whether these actors regard this as more important than the stakes of the political dispute itself. This point is central to our argument. This is because it is plausible to assume that the stakes of the political dispute will be higher when more issues are politically contested and the political landscape is more polarized. Indeed, as the existing studies show, voter polarization makes them more likely to view politics as a high stakes competition (Ward & Tavits, 2019). This signifies that in cases of stronger political disagreement (political conflict and/or political polarization), fidelity crises may be more likely to emerge, as in disputes with their opponents, national political leaders may be more willing to vindicate power or advance certain interests beyond constitutional bounds. In the theoretical framework proposed by Voigt (2021), this would imply that the benefits of overstepping the constitution, in terms of achieving political goals, outweigh the costs to these actors resulting from non-compliance. Put differently, increase in political polarization (whether among political elites or the citizens) may raise political stakes. This, in turn, increases the benefits of constitution-violating actions if the latter are meant to achieve other goals salient in the dispute. At the same time, however, the costs of behavior non-complying with the constitution (for example, bringing politicians before a court empowered to hold them accountable for these actions) will remain the same. In effect, with higher levels of political polarization, politicians may be more willing to consider overstepping the constitution.

In line with this view, Whittington (2002) argues that two of the crucial reasons, why the American constitutional system has historically been relatively less prone to crises, are the fact that politics in the United States have often been less polarized than in other countries (giving rise to fewer constitutional fidelity crises), as well as thanks to the common, relatively strong American constitutional culture. He indicates that political actors in the American system accept, to a great degree, the importance of constitutionalism and any disagreements in this area may rather concern its exact meaning and specific requirements. “A shared constitutional culture helps keep political disagreements from expanding into foundational, constitutional disagreements and helps provide the common ground upon which political disagreements can be resolved” (Whittington, 2002, p. 2145). That said, the presence or the endurance of such a constitutional culture is not certain at all, regardless of the geographical context.

Whittington’s reflections on constitutional crises also hint at two other arguments supporting the view that political polarization can drive constitutional non-compliance. Firstly, the political dispute may concern the very functioning of the constitution (transforming an operational crisis into a fidelity crisis). In this case, one side of the dispute will, by definition, have preferences that are not aligned with at least some constitutional rules. Assuming that increasing political polarization within a society would mean an increasing proportion of the society adopting more extreme ideological positions, public consent for violating constitutional rules by political leaders may grow with the level of political cleavages. This, in turn, may lower perceived costs of constitutional violations by politicians acting in non-compliance with the constitution as they may count on their supporters to oppose any attempts to punish such non-complying behavior.³

Secondly, and relatedly, with high levels of political polarization in society, the conflict can be about many issues. Under this scenario, potential violations of constitutions will be bundled with other matters in the dispute. Because for many constitutional issues may not be the most salient ones, their efforts and expectations may be directed towards a different field of political conflict. In this case, citizens may be less sensitive to instances of political elites’ behavior non-complying with constitutions and/or they may be willing to sacrifice this issue for gains on another front. Be as it may, in either case, political polarization would mean a higher probability of political actors’ non-compliance with the constitution. Below, we aim at testing whether this view finds support in the data.

3 Empirical tests

The empirical part of our study, aimed at investigating the potential links between political conflict, political polarization and constitutional compliance, is based on a global sample, limited by data availability to ca. 170 countries (for the baseline models—see the list of countries in “Appendix 1”). The timespan of our analysis covers the period 1975–2020, as dictated by the scope of one of our core sources of data (i.e. the Database of Political Institutions—Cruz et al., 2021).

Drawing on the literature review and theoretical background presented in the previous section, we aim to test the following hypothesis: Higher levels of constitutional non-compliance are more likely to emerge at times of intensified political conflict—in circumstances of highly polarized political landscapes, as well as in highly politically polarized societies. Thus, our empirical strategy involves estimating several variants of panel data regressions associating a measure of the degree of constitutional compliance as the dependent variable with numerous indicators of political polarization as explanans, accompanied by some controls.

Our dependent variable is a novel measure of constitutional compliance (cc_total) originating from the Comparative Constitutional Compliance Database (CCCD), as proposed by Gutmann et al. (2024b). This aggregate measure includes indicators for governments’ compliance with 14 constitutional rules grouped into four legal areas: (1) property rights and the rule of law (encompassing property rights, judicial independence, equality before the law, and rule of law); (2) political rights (freedom of association, freedom of assembly, and the right to form parties); (3) civil rights (free media, free speech, free movement, and religious freedom); and (4) basic human rights (the right to life, freedom from slavery, and protection from torture). The construction of each of these individual indicators follows Voigt’s (2021) conceptualization of constitutional (non-)compliance and involves matching information on de jure constitutional rules from the Comparative Constitutions Project (CCP—Elkins et al., 2021) with data on these rules’ de facto implementation provided by the Varieties of Democracy project (Coppedge et al., 2022). Compliance with a constitutional rule is coded 1 if the given rule is protected both de jure and de facto, 0 if the right is protected de jure, but not de facto (i.e. there exists a de jure–de facto gap), and 0.5 if a constitutional right is not protected de jure, irrespective of the de facto indicator.⁴ Individual scores for these 14 compliance indicators, as well as aggregated measures for the four above-mentioned categories of constitutional rules and an overall compliance indicator across all rules, are calculated using factor analysis for 175 countries over the time period 1900–2020. Advantages of this dataset include not only wide coverage but also that all de jure and all de facto data, respectively, come from the same data source, making it possible to avoid problems of comparability due to differing coding rules and measurement strategies for compliance with different rules.

With regard to our crucial independent variable, i.e., the measure capturing the degree of political polarization (political_polarizaton), we refer to two dimensions: political polarization of the society and political polarization of the political landscape. The first is proxied by the degree to which the society is polarized into antagonistic, political camps (soc_polariz) and the data is drawn from the Varieties of Democracy Dataset (Coppedge et al., 2022). The second is measured by a set of three variables based on data from the Database of Political Institutions (Cruz et al., 2021): polarization between the executive party and the principle parties of legislature (el_polariz), fractionalization of the government (gov_frac), as well as fractionalization of the opposition (opp_frac).⁵ Pairwise correlations between these variables are presented in “Appendix 2”.

In addition, we include two control variables relating to the robustness of the civil society in a given country (civil_society), as well as the country’s degree of international autonomy (the extent to which a state is autonomous from the control of other states with respect to the conduct of its foreign policy—int_autonomy), as potential additional drivers of constitutional compliance suggested by the literature (data for both variables originates from the Varieties of Democracy Dataset—Coppedge et al., 2022). A more detailed description of all variables used in this study, together with their data sources, is provided in “Appendix 3”. Table 1 reports the descriptive statistics for our main variables of interest.

Table 1 Descriptive statistics

Our empirical strategy is based on estimating a variety of panel data models. We start with a fixed-effects (FE) model characterized by the following functional form:

$$\begin{aligned} cc\_total_{i.t} & = \varphi political\_polarization_{i,t} + \beta civil\_society_{i,t} \\ & \quad + \delta int\_autonomy_{i,t} + \alpha_{i} + \gamma_{t} + \varepsilon_{i,t} \\ \end{aligned}$$

(1)

where ${\alpha }_{i},{\gamma }_{t},{\varepsilon }_{i,t}$ denote the country fixed-effects, time fixed-effects and error term respectively.⁶ The core advantage of executing fixed-effects models is the opportunity to control for country- and year-specific effects, in order to mitigate the problem of omitted variables. In an alternative specification, we control for constitution-system-⁷ and year-specific effects.

In addition, we conduct Arellano-Bond linear dynamic panel estimations (Arellano & Bond, 1991), given that including the lagged dependent variable provides us with the possibility to capture the dynamic effect in our model. Crucial changes in constitutional compliance are usually effects of gradual improvements or weakening, not startling highs and lows (Lewkowicz & Lewczuk, 2023). As far as these models are concerned, we use the following set of equations in first-differences with one-period lagged levels of the dependent variable:

$$\begin{aligned} cc\_total_{i,t} & = \gamma cc\_total_{i,t - 1} + \varphi political\_polarization_{i,t} \\ & \quad + \beta civil\_society_{i,t} + \delta int\_autonomy_{i,t} + u_{i} + \varepsilon_{i,t} \\ \end{aligned}$$

(2)

where ${u}_{i}$ stands for the unobserved time-invariant country effect, and ${\varepsilon }_{i,t}$ is the error term. Importantly, using robustness checks to verify the validity of our results as well as enrich them, we examine the outcome of the regressions for the dependent variable taken as a moving average of the levels of $cc\_total$ in a given country in periods $t$, $t+1$ and $t+2$. This is intended to capture the underlying dynamics of the relatively slowly evolving phenomenon of constitutional compliance.⁸ Moreover, we include one-period and three-period lags of our variables related to political polarization (soc_polariz, el_polariz, gov_frac and opp_frac) to account for the fact that they may affect constitutional compliance with a certain delay. Last but not least, we estimate all our models additionally on subsamples of the data, grouping countries based on their crucial political and economic characteristics. Specifically, we classify them according to their regime type (democratic/non-democratic), based on data for electoral democracy from the Bjørnskov–Rode Regime Dataset (Bjørnskov & Rode, 2020), as well as according to income levels (low-income, lower-middle-income, upper-middle-income, and high-income economies, as proposed by the World Bank, 2022).⁹ Division by regime type in our case means that we select all democratic or non-democratic countries (classified according to the electoral aspect) for each year from the period 1975–2020, whereas classification by income levels is made as of 2022.

Before moving to our results, what needs to be stressed is that our findings come with an important caveat. More specifically, notwithstanding the fact that we control for various types of fixed effects or attempt to estimate dynamic panel models, what we show here should be read as conditional correlations and not causal effects. This is for two basic reasons. Firstly, our models are likely to suffer from omitted-variable bias. Secondly, while using a lagged dependent variable among explanatory variables may solve some endogeneity problems, it is likely to introduce other identification issues (Bellemare et al., 2017).

Tables 2, 3 and 4 present estimation results for the baseline specifications: Table 2 and 3 refer to the models including country and year or constitution-system and year fixed effects, while Table 4 refers to the Arellano-Bond dynamic estimations. For the sake of clarity, tables with subsequent estimation results have been shifted to “Appendix 4”. Tables 5, 6 and 7 report the results for the regressions accounting for potential delayed effects. Tables 8, 9 and 10, in turn, present the output for country groups.

Table 2 Political polarization and compliance with the constitution (country and year fixed-effects)

Table 3 Political polarization and compliance with the constitution (constitution-system and year fixed-effects)

Table 4 Political polarization and compliance with the constitution (dynamic approach)

The results of our models imply a stable, statistically significant and negative association between political polarization of the society and constitutional compliance, which is in line with our expectations based on the theoretical and literature background presented earlier. As such, these findings conform with the claim that deeply politically polarized societies are more vulnerable to constitutional misbehavior of their governments.¹⁰

When we turn to the polarization of the political landscape, we observe more diverse results. In Table 2, we find evidence of statistical significance of most of the variables included to capture this context—el_polariz, gov_frac and opp_frac, but the values of the coefficients are very small—two to three orders of magnitude lower than the ones for soc_polariz. In subsequent regressions (Table 3, 4, 5, 6 and 7), however, in many cases, these variables either fail to maintain their significance or the signs of their coefficients change. Based on the full set of results, therefore, we cannot speak of any coherent findings in this respect, neither on the global scale, nor for specific groups of countries.¹¹

All in all, our findings seem to be partly in line with the hypothesis stated in this paper and suggest that indeed intensified political conflict in highly politically polarized societies may be positively related to constitutional non-compliance. Interestingly, our findings tend to indicate that the relationship between adherence to the constitution and polarization within society seems to be more robust than the relationship between adherence to the constitution and degree of polarization of the political landscape. This implies deeper roots of the relationship between political conflict, polarization and constitutional compliance—reaching beyond verbal skirmishes between political actors or the distribution of seats in the parliament. Our findings indicate that public feelings, and polarization within the society in particular, might be more important in this respect, suggesting that once the social debate becomes too heated, political elites can be emboldened to act against the constitution. As we explained, the increase in political polarization in the society may raise political stakes and, in turn, lead to an increase in the benefits of constitution-violating actions if the latter are meant to achieve other goals salient in the dispute. To the extent that this is accompanied with the costs of non-compliance being unchanged, politicians may be more willing to choose to overstep the constitution. In addition, the perceived costs for politicians acting in non-compliance with the constitution may also decrease in the case of increased political polarization of the society, as they may count on their supporters to oppose any attempts to punish such non-complying behavior. Bundling constitutionally questionable behavior with other issues may also be of relevance.

Finally, as far as control variables are concerned, robustness of the civil society and the degree of international autonomy, generally, prove to be relevant for constitutional compliance in our regressions. Throughout all conducted estimations, a stronger civil society is positively and significantly associated with higher levels of constitutional compliance. Similarly, according to the majority of the obtained results, higher degrees of international autonomy are positively significantly related to higher levels of constitutional compliance, with an exception of a few negative coefficients in the dynamic setting and an interesting finding of a negative association between these variables for non-democratic countries. These observations, including the latter, are in line with the existing literature, confirming once again the potential role of foreign actors and civil society organizations in constraining government actors from overstepping constitutions.

4 Conclusions

In this paper, we found that political polarization of a society is associated with a constitutional failure demonstrating itself in the divergence of constitutional practice of government actors from the constitution text. We presented the underlying economic approach to compliance with constitutions and brought forward the political science concept of constitutional fidelity crises to explain the missing link between political conflict/political polarization and constitutional compliance. Even if only in the correlational sense, the results of our empirical study conducted for the global sample are broadly consistent with the argument postulating the existence of such a relationship; however, only with regard to political polarization of the society. The identification of specific mechanisms behind these results remains a future (promising) outlook. Last but not least, the output of our empirical study is not detached from the currently widely discussed events, as it addresses, to some extent, the question about the reasons behind the latest constitutional crises or major controversies in countries like Hungary, Poland, Turkey, Venezuela, and the United States, which were characterized by very strong political polarization of the society in recent years, as indicated in the Varieties of Democracy Dataset (Coppedge et al., 2022).

Appendices

Appendix 1

List of countries included in the study

Afghanistan, Albania, Algeria, Angola, Argentina, Armenia, Australia, Austria, Azerbaijan, Bahrain, Bangladesh, Barbados, Belarus, Belgium, Benin, Bhutan, Bolivia, Bosnia and Herzegovina, Botswana, Brazil, Bulgaria, Burkina Faso, Burundi, Cambodia, Cameroon, Canada, Cape Verde, Central African Republic, Chad, Chile, China, Colombia, Comoros, Congo, Congo (Democratic Republic), Costa Rica, Cote d'Ivoire, Croatia, Cuba, Cyprus, Czech Republic, Denmark, Djibouti, Dominican Republic, Ecuador, Egypt, El Salvador, Equatorial Guinea, Eritrea, Estonia, Eswatini, Ethiopia, Fiji, Finland, France, Gabon, Gambia, Georgia, Germany, Ghana, Greece, Guatemala, Guinea, Guinea-Bissau, Guyana, Haiti, Honduras, Hungary, Iceland, India, Indonesia, Iran, Iraq, Ireland, Italy, Jamaica, Japan, Jordan, Kazakhstan, Kenya, Korea (North), Korea (South), Kuwait, Kyrgyzstan, Laos, Latvia, Lebanon, Lesotho, Liberia, Libya, Lithuania, Luxembourg, Madagascar, Malawi, Malaysia, Maldives, Mali, Malta, Mauritania, Mauritius, Mexico, Moldova, Mongolia, Morocco, Mozambique, Myanmar, Namibia, Nepal, Netherlands, New Zealand, Nicaragua, Niger, Nigeria, North Macedonia, Norway, Oman, Pakistan, Panama, Papua New Guinea, Paraguay, Peru, Philippines, Poland, Portugal, Qatar, Romania, Russia, Rwanda, Saudi Arabia, Senegal, Sierra Leone, Singapore, Slovakia, Slovenia, Solomon Islands, Somalia, South Africa, Spain, Sri Lanka, Sudan, Suriname, Sweden, Switzerland, Syria, Taiwan, Tajikistan, Tanzania, Thailand, Timor-Leste, Togo, Trinidad and Tobago, Tunisia, Turkey, Turkmenistan, Uganda, Ukraine, United Arab Emirates, United Kingdom, United States, Uruguay, Uzbekistan, Vanuatu, Venezuela, Vietnam, Yemen, Zambia, Zimbabwe.

Appendix 2

Pairwise correlation between soc_polaric, el_polariz, gov_frac and opp_frac

	soc_polariz	el_polariz	gov_frac	opp_frac
soc_polariz	1
el_polariz	− 0.1557	1
gov_frac	− 0.1189	0.5872	1
opp_frac	− 0.1679	0.3517	0.5979	1

1. Source: own elaboration

Appendix 3

Description of variables included in the study with data sources

Variable name	Description, based on the source of data	Data source
cc_total	An indicator of government’s compliance with the constitution (de jure–de facto gap) aggregated using factor analysis for 14 constitutional rules grouped in four areas: (1) property rights and the rule of law; (2) political rights; (3) civil rights; and (4) basic human rights. Compliance with an individual constitutional rule is coded 1 if the given rule is protected both de jure and de facto, 0 if the right is protected de jure, but not de facto, and 0.5 if a constitutional right is not protected de jure, irrespective of the de facto indicator	Gutmann et al. (2024b)
soc_polariz	Continuous variable envisaging the extent to which political differences affect social relationships beyond political discussions (original variable name: v2cacamps). The higher the value, the more society is polarized into antagonistic, political camps. Continuous version of the variable was created from the ordinal variable using the measurement model. The ordinal variable envisages the responses to the following question “Is society polarized into antagonistic, political camps?”. Possible responses: 0: Not at all. Supporters of opposing political camps generally interact in a friendly manner 1: Mainly not. Supporters of opposing political camps are more likely to interact in a friendly than a hostile manner 2: Somewhat. Supporters of opposing political camps are equally likely to interact in a friendly or hostile manner 3: Yes, to noticeable extent. Supporters of opposing political camps are more likely to interact in a hostile than friendly manner 4: Yes, to a large extent. Supporters of opposing political camps generally interact in a hostile manner	Coppedge et al. (2022)
el_polariz	Maximum polarization between the executive party and the four principle parties of the legislature (original variable name: polariz). The index is the maximum difference between the chief executive’s party’s value and the values of the three largest government parties and the largest opposition party. The index is zero if the chief executive’s party has an absolute majority in the legislature	Cruz et al. (2021)
gov_frac	The probability that two deputies picked at random from among the government parties will be of different parties (original variable name: govfrac). Equals NA if there is no parliament. If there are any government parties where seats are unknown (cell is blank), govfrac is also blank. No parties in the legislature results in NA	Cruz et al. (2021)
opp_frac	The probability that two deputies picked at random from among the opposition parties will be of different parties (original variable name: oppfrac). Equals NA if there is no parliament. If there are any opposition parties where seats are unknown (cell is blank), oppfrac is also blank. No parties in the legislature results in NA	Cruz et al. (2021)
civil_society	Continuous variable envisaging how robust civil society is (original variable name: v2xcs_ccsi). The higher the value of variable is, the more robust is civil society in a given country. The index is formed by taking the point estimates from a Bayesian factor analysis model of the indicators for governmental control over entry and exit of civil society organizations (CSO) into public life, CSOs’ repression and CSO participatory environment	Coppedge et al. (2022)
int_autonomy	Continuous variable envisaging the extent to which the state is autonomous from the control of other states with respect to the conduct of its foreign policy (original variable name: v2svinlaut). The higher the value, the more autonomous the state is. Continuous version of the variable was created from the ordinal variable using the measurement model. The ordinal variable envisages the responses to the following question “Is the state autonomous from the control of other states with respect to the conduct of its foreign policy?”. Possible responses: 0: Non-autonomous 1: Semi-autonomous 2: Autonomous	Coppedge et al. (2022)

1. Source: own elaboration

Appendix 4

See Tables 5, 6, 7, 8, 9, 10.

Table 5 Political polarization and compliance with the constitution (dependent variable as a moving average of cc_total in t, t + 1 and t + 2)

Table 6 Political polarization and compliance with the constitution (one-period lags of soc_polariz, el_polariz, gov_frac and opp_frac)

Table 7 Political polarization and compliance with the constitution (three-period lags of soc_polariz, el_polariz, gov_frac and opp_frac))

Table 8 Political polarization and compliance with the constitution by political and income groups (country and year fixed-effects)

Table 9 Political polarization and compliance with the constitution by political and income groups (constitution system and year fixed-effects)

Table 10 Political polarization and compliance with constitution (dynamic approach) by political and income groups