Measuring Political Institutions in the Long Run: A Latent Variable Analysis of Political Regimes, 1810–2018

Abstract

In this article, latent variable analysis is used to construct hybrid measure of political development based on the plausible common variation between objective and subjective indicators of political institutions. For a sample of 167 countries for the period 1810–2018, we chart long-term paths of political development. Our empirical strategy attempts to overcome the existing potential bias in the measures of democracy in the long run by extracting the institutional characteristics of political regimes, voter turnout, expert-based assessments and electoral outcomes into two latent indices of political development that can be compared both across space and time. The evidence reveals the remarkable persistence of multiple peaks in the world distribution of political development and uncovers contrasting long-term trajectories across countries traditionally featured in the political economy literature. Our findings add to the current debate about measurement of democratic backsliding.

1 Introduction

The worldwide spread of democracy in the past thirty years represents a major institutional transformation in history (Diamond, 2010; Fukuyama, 1992; Huntington, 1993; Shin, 1994). The gradual spread of democracy also remains a subject of continuous controversy in the political economy and comparative politics literatures. Although many scholars agree that the transition from autocratic regimes into democratic political systems nonetheless conveys many implications for socioeconomic outcomes, many disagreements arise on how to quantify and measure democracy into coherent indicators. State of democracy can be viewed either as a dichotomous measure of political institutions where regimes are divided between democracies and non-democracies, or as a continuous measure where democracy itself is a question of degree and typology (Knutsen et. al. 2023). Moreover, democracy can be viewed as a formal political regime defined by constitutions, laws and electoral systems, and representative institutions (de jure) or as a factual implementation of such arrangements, thus enabling access to collective action (de facto).

This article sets to quantify the dynamics of political development for a panel of 167 countries from the dawn of the 19 th century to the present and combine them into two latent original indices. It provides a step towards a more conclusive measurement of political institutions in the long-term perspective. The article discusses the trajectories of formal checks and balances and the non-elites collective-action capacity for 167 countries in the period 1810–2018 rather than focusing on the democracy as a dichotomous political institution. Such a step provides a necessary leap forward to quantify the institutional characteristics of regimes and better under institutional diversity among the political regimes and their performance over time. Our approach focuses on the objective characteristics of political regimes as well as more subjective, expert-based assessments of the state of democracy using the principal component analysis to extract these two key dimensions of democracy from the underlying components by Marshall et al. (2013), Vanhanen (2000, 2003) and Coppedge et al. (2021).

In our approach, the formal aspect of political regimes captures the institutional diversity of political regimes as specified by formal institutions, broadly delimited between autocratic and democratic regimes. The collective-action aspect captures the factual inclusiveness and contestation as the fundamental concepts of democracy and is essential in understanding the degree of democracy enabled by the existing formal checks and balances (Dahl, 1972; Coppedge et al. 2011). Our method also overcomes the definition of democracy as a dichotomous state of political institutions. It considers democracy as a continuous variable that is a question of degree, not of kind. Using a principal component analytical approach for 167 countries across various sub-periods during the years 1810–2018, our latent trajectories of the state of democracy provide a more nuanced hybrid insight into the long-run patterns of political development that is not reflected in the existing arbitrarily aggregated and potentially inconsistent measures, and best combines both objective, input-based indicators as well as more subjective, expert-coded assessments that extracts the maximum variation from the pooled set of indicators.

The rest of the paper is organized as follows. Section 2 provides a critical survey on measuring political institutions. Section 3 outlines the building blocks of our approach. Section 4 discusses the results and implications of our measures. Section 5 concludes. In appendix, we report results concerning paths of political development with a focus on the United States and Mexico as well as major turning points.

2 Conceptual Framework

2.1 The Role of Political Institutions

Several definitions of institutions have been proposed in the existing literature but few of them have been quantified into synthetic measures of institutions. In the seminal contribution, North (1991) originally defined institutions as “humanly devised constraints that structure political, economic and social action,” (p. 97) that can be divided into formal rules (constitutions, laws, property rights, electoral systems), informal constraints (customs, tradition, codes of conduct) and enforcement mechanism. In the political economy literature, institutions are devised to establish order and reduce uncertainty in economic exchange (Greif, 1989).¹ Furthermore, Acemoglu and Robinson (2012) distinguished between extractive and inclusive political and economic institutions. Inclusive economic institutions are those that facilitate and encourage broad-based participation in economic opportunities that allows people to make the best use of their talents and skills, and to pursue the choices they wish. Such institutions feature secure property rights, an unbiased system of law and order, and a provision of public services that foster a level-playing field in economic exchange and contract enforcement. Extractive economic institutions are designed by the elite to extract economic resources from non-elites. Economic institutions do not exist in a vacuum but are support by the set of political institutions (Acemoglu and Robinson 2006a; Acemoglu et al. 2005a). Extractive political institutions concentrate power in the hands of the few without executive constraints, rule of law and political accountability whereas inclusive political institutions allow broad-based political competition and participation by placing constraints, checks and balanced on the executives.

Several competing views of the origins of political institutions have been proposed. For example, Boettke et al. (2008) identified (i) foreign-induced exogenous institutions, (ii) indigenously introduced exogenous institutions, and (iii) indigenously introduced endogenous institutions. The first type of such institutions is imposed from above by unelected foreign officials, groups, or international organizations. The second and third type of institutions evolve by a spontaneous process where the second type is created by the ruling elite whilst the third type is created by the non-elites. Each type of institutions evolves through varying degrees of path dependence and stickiness where the endogenous institutions are the stickiest of all and can resist the effects of any new rules for a very long time.²

The historical evidence clearly underlines that political institutions designed to structure the economic activity can either persistent over long periods or be overhauled during the major conflict episodes. Insightful empirical evidence on the economic effects of political institutions clearly demonstrates the importance of institutions but tells little on how to measure the institutions more precisely (Robinson, 2013; Shirley, 2013; Voigt, 2013). As such, measures of institutions should refer to specific institutions because overly aggregate measures are too broad to contain any meaningful information. Furthermore, measurements of institutions should primarily be inferred from objective rather than subjective indicators. More specifically, institutions should be measured as formally specified in legislation (de jure) and as factually implemented (de facto).

2.2 Measuring Political Institutions

Since any measure of political institutions is a latent variable that cannot be directly observed, the agreement among political economy and comparative political science scholars on how to measure political institutions is subject to enduring controversy (Boese, 2019; Boswell & Corbett, 2021). Scholars have developed numerous measures on the appropriate measurement of democracy and typically disagree on the components of democracy, unit of the measure, and how different components can be aggregate into a single summary index. Selecting a single measure to operationalize political development often forces researchers to omit potentially useful information in alternative measures and data whilst subtle differences in the array of components, measurement and aggregation techniques do affect the substantive results on the link between democracy and various socioeconomic outcomes (Adcock & Collier, 2001; Casper & Tufis, 2002; Collier & Adcock, 1999; Elkins, 2000).

The source of disagreement on the measurement of democracy primarily stems from the content and features of political development. In the original formulation, Schumpeter (1950) presented a minimalist definition of democracy where representatives are required to compete for popular vote. Dahl (1972) introduced the concepts of contestation and inclusiveness as the fundamental concepts of democracy. The former captures the competition among the representatives with free and fair elections for the legislature and accountable executive either directly to the people or to the elected legislature. The latter captures the popular voting participation and requires the extension of suffrage to the adult population for the country to qualify as democratic (Lipset, 1959; Sartori, 1987; Boix, 2003, Paxton et al., 2003, Mainwaring et al., 2007).

2.2.1 Measures of Political Development

Numerous measures of democracy that have been proposed on the basis of selected criteria and numerical forms which can be classified into two major types. First, the dichotomous measure of democracy designates societies as either democratic or non-democratic based on the number of pre-determined criteria that reflect the extent to which political institutions facilitate political competition among representative and participation in the popular vote. In the dichotomous setting, democracy is a binary variable. Many scholars have proposed the dichotomous measurement of democracy (Sartori, 1987; Huntington, 1991; Alvarez et al., 1996; Przeworski et al., 2000; Golder, 2005; Cheibub et al., 2010; Skaaning et al., 2015). And second, alternative measure denote democracy either as an ordinal measure (Collier & Levitsky, 1997; Diamond, 2002; Linz & Stepan, 1996), continuous measure (Bollen & Jackman, 1989; Cutright, 1963; Elkins, 2000), single-dimensional measure (Bollen & Grandjean, 1981) or multi-dimensional measure (Bollen & Paxton, 2000, Gates et al., 2006, Gerring et al., 2009). Another source of disagreement concerns the measurement of various components of democracy. In the broadest form, democracy can be measured on the scale through the aggregation of pre-determined components (Coppedge & Reincke, 1991; Gledisch & Ward, 1997; Munck & Verhuillen, 2002) or as a latent variable (Coppedge et al., 2008; Bollen, 1980; Trier & Jackman, 2008; Pemstein et al., 2010; Miller 2011) where multivariate techniques are used to extract the components into the measure of democracy. Earlier studies have shown that different measures directly affect the results on the various effects of democracy (Bogaards, 2010; Boix, 2011; Boix & Stokes, 2003; Vreeland, 2008).

Several attempts have been proposed to measure democracy. The first and perhaps the most comprehensive such attempt is the Polity IV Index (Marshall et. al. 2013) where the institutional characteristics of political regimes such as the patterns of democratic and autocratic authority are assessed for 167 countries with populations greater than 500,000, covering the years 1800–2013. The indicators used to construct the measures of democracy and autocracy are both accessible and well-document are created from five expert-coded categories: (i) competitiveness of executive recruitment, (ii) openness of executive recruitment, (iii) executive constraints, (iv) regulation of participation, and (v) competitiveness of the participation. The summary measure of democracy is denoted on the ordinal scale ranging from −10 to 10 where higher values indicate greater democracy. Polity IV Index captures the qualitative de jure institutional characteristics of political regimes delegated by formal institutional framework such as laws, constitutions and electoral systems, and has been used extensively in empirical applications on the link between political institutions and development outcomes (Almeida & Ferreira, 2002, Glaeser et al., 2004, Rodrik & Wacziarg, 2005, Mobarak, 2005, Jones & Olken, 2005, Persson & Tabellini, 2008, Yang, 2008, Acemoglu et. al. 2009, Klomp & De Haan, 2009, Evans, 2009, Papaioannou & Van Zanden, 2015).

The second attempt to measure the patterns of democracy and autocracy in the long run is the index of democracy by Vanhanen (2000, 2003). The index covers 187 countries in the period 1810–2000 and presents one of the first attempts to operationalize Dahl's (1972) contestation and inclusiveness concepts as the fundamental characteristics of democracy. In comparison with Polity IV Index, Vanhanen's index of democracy comprises two quantitative measures reflecting the public contestation and the right to participate in the collective action. First, the degree of political competition comprises the proportion of votes won by smaller political parties in parliamentary and presidential election, and indicates the degree of political competition in a given political system. And second, the degree of political participation denotes the proportion of the adult population that actually voted in these elections and is used as a measure of political participation which reflects the ability of the population to engage in various forms of collective action. The summary measure of democracy is constructed as an equally weighted average of both indices in the range between 0 and 100 where higher values indicate greater democracy. Vanhanen's index of democracy captures the quantitative de facto institutional characteristics of political regimes that indicate the ability of the adult population to engage in various forms of collective action and has been used extensively in empirical applications on the link between political institutions and socioeconomic outcomes (Bjørnskov, 2010; Chowdhury, 2004; Jakobsen & De Sousa, 2006; Landman, 1999; Li, 2005; Neumayer, 2002; Pinto & Timmons, 2005).

Additional attempts have sought to quantify the degree of democracy across political systems as a dichotomous measure where the degree of democracy is denoted in a binary fashion either as zero for non-democracies and one for democracies based on the specified number of criteria. Alvarez et al. (1996) classified the political regimes across 141 independent countries in the postwar period (1950–1990) and constructed the binary measure of democracy if (a) the executive is directly or indirectly elected in the popular election, (b) if the legislature is elected in the popular vote, and (c) if there is more than one political party. A similar dichotomous measure has been constructed by Przeworski et. al. (2000) for 135 countries in the period 1950–1990 based on executive and legislative election in the popular vote, and the degree of party competition. Covering 219 countries in the period 1800–2007, Boix et al. (2013) defined a country as democratic if (i) the executive is directly or indirectly elected in popular elections and is held accountable either to the legislature or directly to voters, (ii) if the legislature is chosen in free and fair elections, and (iii) if a majority of adult men has the right to vote. Although such approach captures the relevant dimensions of democracy, too much information is being lost by compressing the degree democracy to a single dichotomous measure.

Another strand of literature pursued an alternative approach by measuring democracy either on continuous or ordinal scale rather than as a dichotomous reflection of political institutions. The first and most popular of these measures of democracy is the index of political rights and civil liberties by Freedom House (2013) which covers 195 countries and 14 disputed territories from 1972 onwards. Indicators of political rights comprises four categories: (i) electoral process, (ii) political pluralism and participation, and (iii) functioning of government. Similarly, the indicators of civil liberties are composed of four ordinal measures, notably (i) freedom of expression and belief, (ii) associational and organizational rights, (iii) rule of law, and (iv) personal autonomy and individual rights. The index is defined on the ordinal scale ranging from 1 to 7 (Gastil, 1985, 1990) where higher scores indicate fewer political rights and civil liberties, and has been used extensively in cross-country empirical research (Barro, 1996; Boone, 1996, Gwartney et al., 1999; Knack & Keefer, 1995; Rodrik, 2000; Tavares & Wacziarg, 2001; Baum & Lake, 2003; De Haan & Sturm, 2003, Acemoglu et. al. 2005b). The scores to each measure are derived from expert-based evaluation which has been subject to the criticism (Bollen, 1990) emphasizing statistical bias of unknown direction stemming from subjective evaluation and the measurement error inherent in the absence of quantitative basis. Such ordinal measures largely focus on post-WW2 measurement of democracy and civil rights whereas a long-run ordinal measure of democracy has been constructed by Bowman et al. (2005) for 5 Central American countries (Costa Rica, El Salvador, Honduras, Guatemala, and Nicaragua) for the period 1900–1999 based on in-depth analyses of primary sources.

More recently, several ambitious research programs have been launched to improve the measurement and reliability of democracy and political development indicators. Against this backdrop, Coppedge et al., (2016, 2019, 2021, 2023) developed a series of high-level democracy indicators based on the large array of lower-level indicators of democracy and institutional development using country-level expert assessments translated into a Bayesian item response model for a large number of countries from 1789 up to the present day. The dataset has been hailed the most elaborate and granular series of indices capturing political development over long period of time (Hegedüs, 2020) in spite of several critiques emphasizing coder-level biases in the assessment of political regimes and state of democracy stemming from overly subjective assessments (Weidmann, 2023) influenced by the recent events (Soroka et al., 2019) which may invoke the gradual abandonment of the pluralist conceptualisation of democracy (Wolff, 2023). To partially fill a void in the literature, Pemstein et al. (2010) synthesized a new measure of democracy for all countries since 1946 onwards by leveraging the measurement efforts of numerous scholars. More specifically, they address potentially large uncertainty in the level of democracy by estimating the posterior distribution of scores alongside the 95 percent confidence intervals that allows research to assess the uncertainty of estimates more closely.

2.2.2 Problems of Measurement Error and Aggregation

Although a significant amount of time and space have been devoted to measuring democracy, each measure is subject to varying degrees of measurement error and aggregation bias. The lack of agreement on the precise measurement of democracy can be summarized into three key research dilemmas. First, the nature of the underlying scores on democracy is debatable since many indices are constructed in seemingly arbitrary fashion with little regard to the justification of the underlying measure. Second, the validity of the constructed measures of democracy is controversial. Coppedge and Reinicke (1991), Bollen (1993) and Gleditsch and Ward (1997) discussed extensively validity issues involved in the democracy indices arising from arbitrary assignment of original component values to the underlying indicators, casting considerable doubt on the reliability and measurement of scale in the additive indices. Third, the aggregation of scores from the principal indicators as a meaningful and non-arbitrary approach to derive a summary index of democracy is subject to discussion. Such bias can cause non-trivial deficiencies to the overall measures of democracy which can be readily observed through the lumpiness of distribution and arbitrary system of weights used to construct the summary index. And fourth, the underlying indices of democracy address the measurement error in multiple ways. Existing measures of democracy assume almost deterministic relationship between various components and the aggregate measure of democracy. Ignoring the fact that each of these components is imperfectly measured with substantial and non-random measurement error, can undermine the validity of the index and render the conclusions biased, implausible and inconsistent. A poorly measured variable also causes the attenuation of underlying responses for the outcomes of interest as a result of errors-in-variables that underestimate the true coefficients (Karni & Weissman, 1974; Hausman & Taylor, 1981; Aigner et al., 1984; Griliches & Hausman, 1986; You & Chen, 2006).

Three major steps forward have been to overcome the problems of measurement error and aggregation bias using the multivariate techniques and modelling democracy as a latent variable. The first such attempt was by Trier and Jackman (2008), using ordinal response model to improve the reliability of the scale of Polity IV measure of democracy. Their approach relied on the inter-relationships between Polity indicators to obtain the scores to weigh the contribution of each indicator to the total score appropriately. It uses item-identification parameters to embed the level of democracy as an unknown parameter. Such an approach is ought to deliver both point estimates and the distribution of democracy scores across countries. Trier and Jackman also called for greater number of democracy indicators to construct a better latent measure of democracy. The second such attempt was done by Pemstein et al. (2010), using Bayesian latent variable approach to synthesize an original measure of democracy from ten existing extant scales, to address the measurement error in existing indices and to derive a new latent measure of democracy for 191 countries for the period 1946–2012. Their measure of democracy (Unified Democracy Score) is adjusted for the uncertainty of the score from the extant scales and at least as reliable as the most credible component measure and directly comparable across countries.

The third attempt and most comprehensive attempt so far was by Földvari (2014a, 2014b). It adopts two latent variable methods with measurement error and extracts the components of Polity IV and Vanhanen Index of Democracy to analyse the relationship between de jure and de facto political institutions for 134 countries in the period 1820–2000, under the assumption that each of these components is driven by a common latent factor. Following the original definition of democracy by Dahl (1972), Földvari’s approach was the first one so far to address the selectivity bias in the underlying components driven by the change in the number of available countries. It showed that a substantial fraction of observed relationship between the measures of de jure and de facto political institutions is explained by sample selection bias. Moreover, it allowed for direct comparison of the estimated latent de jure and de facto measures of democracy across a large number of countries and does not confine the temporal coverage to post-WW2 period.

2.2.3 The Recent Debate on Measuring Democratic Backsliding

An intense debate has emerged in the last two years concerning the possibility that current available indicators misestimate democratic declining or backsliding in Europe and in North America since the 2010s, in particular Varieties of Democracy (Knutsen et al., 2023; Little & Meng, 2024). Although available indicators are strongly correlated, Varieties of Democracy is perceived to outperform Polity IV and Freedom House Index with respect to the underlying definition, measurement scale, and aggregation procedure (Boese, 2019). Unsurprisingly, the debate has focused primarily on Varieties of Democracy. Scholars point out to three areas of contention. One concerns the underlying conceptual ambiguity and pluralism of democracy and the role of subjective perceptions and views in characterizing democratic declining (Boswell & Corbett, 2021; Wolff, 2023). A second dispute reflects potential biases in coding the data or in shaping aggregation procedures, thus suggesting alternative methodologies (Weidmann 2023; Weitzel et al., 2024). Finally, scholars hold different views about the effectiveness of democratic weakening, in particular from the viewpoint of long-term trends in political institutions (Treisman, 2023).

3 Measuring Long-run Political Development: Latent Variable Approach

3.1 Our Approach

Our article identifies the common factors of political institutions in the long-run perspective for a large number of countries following the original distinction between formal characteristics and the de facto elements of political institutions. It suggests how formally specified checks and balances differ from the de facto capacity of non-elites and civil society to engage in various forms of collective action. Using the principal component approach, measures of formal checks and balances and non-elites collective action capacity are extracted from the underlying components that capture a) the formal characteristics of political regimes, and b) factual ability of adult population to contest and participate in the given regime. Principal component scores are obtained for 167 countries from the dawn of nineteenth century to the present in a continuous measure without fixed end-points which allows us to chart long-term contrasting paths of political development and explain the contrasting trajectories of democracies and non-democracies across space and time.

Adopting such distinction as the backbone of our approach allows us to identify the differences in the substantive content of political institutions on one hand, and the factual implementation of such institutions in broadening the ability to engage in various forms of collective action. Such a distinction is crucial for understanding the long-run patterns and world distribution of democracy that might not be inferred from short-run time-series in existing latent measures of democracy that usually start at the onset of post-WW2 period.

Using the data on the formal institutional characteristics of political regimes from Polity IV (Marshall & Gurr, 2020), the data on voter turnout and composition of parliaments (Vanhanen, 2000, 2003) and more recent generation of Varieties of Democracy (Coppedge et al., 2021) indicators of electoral and liberal democracy, allows us to identify the long-run paths of formal checks and balances and strength of civil society for a large number of countries from the existing scales similar to Trier and Jackman (2008) and Pemstein et al. (2010) but with a broader temporal coverage from 1810 to 2018. The approach proposed in this article follows the factor analytical model where hybrid measures of political development combining both objective indicators and subjective perception are extracted from the underlying Polity IV, Vanhanen, and V-DEM components. In addition to Földvari (2014a, 2014b), the measurement error and selectivity bias in the latent democracy score is addressed by constructing time-varying samples for multiple benchmark years, rather than sample selection method, to facilitate country coverage and partially address the heterogeneous distribution of democracy arising from omission bias in given years and data availability. In comparison with existing attempts by Freedom House and more recent democracy indices (Economist Intelligence Unit, 2014), our proposed latent measures of political development (i) are constructed on the basis of long-run patterns rather than short-run deviations which provides a clear and more nuanced view on the long-run evolution of democratic development and its reversals, and at the same time, (ii) avoids the excessive use of subjective measures and the country-specific coder bias in attempting to construct more reasonable measures of political institutions in the long run.

3.2 Political Institutions as Latent Variables: Principal Component Analysis

Our primary goal is to construct latent measures of political development from the extant scales measuring the formal and factual characteristics of political regimes and the actual outcomes in the long-run perspective. Therefore, let $\mathbf{Z}=\left\{{Z}_{1,i,t},{Z}_{2,i,t},\dots {Z}_{p,i,t}\right\}$ be an observable vector of political variables with an unknown number of components denoted by $p$ for a given number of countries $i=\text{1,2},\dots N$ and for the time period $t=\text{1,2},\dots T$. The vector of observable characteristics of political regimes has a country-specific mean $\mu ={N}^{-1}{\sum }_{i=1}^{N}{Z}^{\left(i\right)}$ and a singular variance $\Sigma ={m}^{-1}{\sum }_{i=1}^{m}{\left({Z}^{\left(i\right)}-\mu \right)\left({Z}^{\left(i\right)}-\mu \right)}^{T}$ and follows a Gaussian distribution $Z\sim {\mathbb{N}}\left(\mu ,\Sigma \right)$.

The vector of observable political regime characteristics is linearly related to $p$-th number of common factors where our aim is to reduce the number of components from observable characteristics into the set of common factor that contain the maximum variation from the observable regime characteristics so that $m<p$ holds.

Let $\mathbf{F}=\left\{{F}_{1},{F}_{2},\dots {F}_{m}\right\}$ denote the number of common factors and let ${\varvec{\upvarepsilon}}=\left\{{\varepsilon }_{1},{\varepsilon }_{2},\dots {\varepsilon }_{m}\right\}$ represent the measurement error that captures the idiosyncratic and transitory regime characteristics not reflected by the common factors. More specifically, the common factor model of regime characteristics is capture by the following system of structural equations:

$$\begin{array}{c}{Z}_{1}-{\mu }_{1}={\lambda }_{11}{F}_{1}+{\lambda }_{12}{F}_{2}+\dots +{\lambda }_{1m}{F}_{m}\\ {Z}_{2}-{\mu }_{2}={\lambda }_{21}{F}_{1}+{\lambda }_{22}{F}_{2}+\dots +{\lambda }_{2m}{F}_{m}\\ \begin{array}{c}\vdots \\ {Z}_{p}-{\mu }_{p}={\lambda }_{p1}{F}_{1}+{\lambda }_{p2}{F}_{2}+\dots +{\lambda }_{pm}{F}_{m}\end{array}\end{array}$$

where the coefficient ${\lambda }_{pm}$ denotes the weight of the $p$-th observable characteristic on the $m$-th unobservable common factor of political development. Henceforth, let the error variance of the idiosyncratic regime characteristics take the following shape:

$$cov\left(\varepsilon \right)=\Psi =\left[\begin{array}{ccc}{\Psi }_{1}& \cdots & 0\\ \vdots & \ddots & \vdots \\ 0& \cdots & {\Psi }_{p}\end{array}\right]$$

where ${\Psi }_{p}$ denotes the idiosyncratic political regime factor not accounted for by the underlying $p$-th observable political development variable. Assuming that the set of common factors $\mathbf{F}$ and idiosyncratic factors ${\varvec{\upvarepsilon}}$ are identically and independent distributed with zero structural covariance restriction $\text{cov}\left({\varvec{\upvarepsilon}},\mathbf{F}\right)=0$, the total variance of the underlying composite indicator of political development is:

$${\sigma }_{ii}={h}_{i}^{2}+{\psi }_{i}$$

where ${\sigma }_{ii}$ is the total variance in the latent unobservable political development variable, ${h}_{i}^{2}$ is the variance shared with other observable variables through the common factor, and can be written in the more extensive form as ${h}_{i}^{2}={\lambda }_{i1}^{2}+{\lambda }_{i2}^{2}\pm \dots {\lambda }_{im}^{2}$, and ${\psi }_{i}$ is the idiosyncratic variance unabsorbed by the common factor but is related to the variability in the observable characteristics. Since $\text{cov}\left({\varvec{\upvarepsilon}},\mathbf{F}\right)=0$ implies that the covariance of the idiosyncratic and common factor is not dependent on any specific trait in any way, the common factors account for most of the composite relationship between the observed variables. It should be noted that the weights used to extract the common factor from the vector of observable characteristics is not arbitrary. Instead, the construction of weights is based on the rotation of the matrix of eigenvalues that yields the optimal factor loadings extracting the maximum possible correlation from the inter-related observable components of political development.

Although such approach does not explicitly guarantee that the principal component are true measures of formal regimes characteristics and collective-action capacity of non-elites, the optimal factor loadings reflect the relative weight and importance of each observable indicator in the common factor and thereupon allow a direct comparison between the constructed latent factor variables and the alternative measures and proxies of political institutions and their development over time. In turn, the relevance and feasibility of the principal components can be discussed and assessed empirically subject to rigorous scrutiny.

Our approach conveys a number of notable advantages compared to the existing indices of political institutions and development. Compared to Coppedge et al. (2021), our scale does not yield fixed end points and does not invoke any particular assumption on what comprises perfect state of autocracy and democracy, which does not compress the variation in both the degree of political development both across and within countries. In addition, no assumption on the upper and lower bound of the score and the rotation of factor loadings achieves a zero-threshold boundary as an easily interpretable boundary separating relatively more inclusive and participatory from more extractive and exclusionary political regimes. In addition, our approach provides a relatively clear and clean aggregation rule where the choice of variable levels is captured and motivated by the underlying theory of political institutions without arbitrary weighing of components. Instead, principal component approach yields the vector of weights based on the maximum possible variation in the observable indicators of political development. By avoiding the adoption of rather maximalist definition of political institutional inclusivity, our approach also considers suffrage requirement in the underlying measure of political development. Without suffrage requirement in the data structure and observable indicators, the notion of inclusivity behind political development can only seldom be discussed. At the same time, by combining suffrage requirement with voter turnout and strength of civil society, our approach emphasizes a stable definition of political development distinguishing between both formal characteristics of checks and balances and the capacity of civil society and non-elites to engage in various forms of collective action. In turn, our hybrid measure of political development extracts the maximum feasible variation from the more objective indicators of political contestation (i.e., indicated by voter turnout and party vote shares in national elections) and more subjective assessments recently espoused by a new generation of democracy indicators (Coppedge et al., 2021). Such hybrid measures validity is reinforced by a high degree of internal consistency since the components potentially susceptible to noise-based variance are supressed or downgraded in the optimal factor loadings. Furthermore, such measure is easily replicable and also consistent with the recent advances and insights in the political economy of development clusters (Besley & Persson, 2011) that bolster the importance of effective and cohesive political institutions that promote common interests, guarantee the provision of public goods and enforce peaceful resolution of conflicts.

3.3 Data and Samples

To extract common factors of political development from the observable indicators, we use the data on the structure and development of political institutions from three different sources. First, we use the data on the quality and structure of political institutions from Varieties of Democracy dataset (Coppedge et al., 2021) where the structure, quality and characteristics of political regimes are assessed across 470 indicators composed into five high-level indices that cover 202 polities for the period 1789–2019. Each indicator is coded independently by at least five country experts based on a series of subjective assessments. Five core indicators are denoted on the limited scale between 0 and 1 and distinguish between electoral, liberal, participatory, deliberative and egalitarian aspect of democracy. For the sake of brevity and data availability, we consider the liberal and electoral component of democracy which is quantified and available from early nineteenth century onward for a large number of polities.

Second, we use the Polity IV data on institutional characteristics of political regime (Marshall et al., 2013) for the period 1810–2000 to extract the de jure component of political institutions. The summary measure of Polity IV index (Polity2 score) is denoted on the range from −10 to 10 where higher values indicate greater extent of institutionalized democracy. The summary index consists of five key components: (i) competitiveness of executive recruitment, (ii) openness of executive recruitment, (iii) executive constraints, (iv) regulation of participation, and (iv) competitiveness of the participation. Each component reflects the substantive content of de jure political institutions for a given number of regimes, and in contrast to Coppedge et al. (2021), it does not include suffrage requirement.

And third, we use the data on political competition and participation from Vanhanen (2000, 2003) for the period 1810–2018 to extract the maximum variance from each component and construct the latent measure of de facto political institutions. The original summary measure is an unweighted average of both indices and is denoted on the scale from 0 and 100 where higher values indicate greater extent of de facto political institutions and greater engagement and ability to engage in various forms of collective action. The index of political competition is constructed as a proportion of votes won by smaller political parties whereas the index of political participation reflects the actual share of the adult population who voted in presidential and parliamentary elections. Both indices are denoted in percentage terms. Compared to Polity IV index, Vanhanen index does not reflect the institutional characteristics of political regimes but the factual implementation of the political democracy in the degree of competition between political parties and in terms of the voter turnout. In the summary index, the aggregate measure of democracy is constructed as an unweighted average of the competition and participation indices since it might be impossible to decide which of the two indices should be attached higher weight.

Using the principal component analysis instead of the arbitrary aggregation procedures allows us to extract the maximum variance from each underlying component and extract it into a single index of de jure political institutions. Such an index conveys four main advantages. First, it overcomes the aggregation bias arising from the arbitrary use of weights upon each component. Such weight structure might not reflect the fundamental characteristics of political regimes in terms of the degree of democracy since some components potentially reflect an opposite pattern from the one found in the other components. In the proposed latent variable framework, we extract the maximum variance in each underlying component and construct a latent measure of de jure political institutions which overcomes the aggregation bias arising from the neglect of the hidden variance of each component in the summary index. Second, the factor approach allows us to explicitly tackle the continuity assumption by removing the arbitrary scale where regimes are ranked on pre-determined scale which can be found in Polity2 score. Imposing a pre-determined limit on the index of political institutions can compress the insightful variation in the de jure characteristics of regimes and hence prevents the key sources of institutional differences from further exploitation. Subsequently, some regimes can be mischaracterized in terms of comparative de jure institutional outcomes. Third, the factor analytical approach allows us to treat de jure political democracy as a continuous variable rather than as a binary outcome. The latter is clearly at odds with the continuity assumption and can potentially misclassify some regime as either democratic or non-democratic when the legislative features of the given regime comply with neither the former nor the latter characteristic. And fourth, the underlying factor loadings provide a more coherent balance between objective indicators on voter turnout, structure of national parliaments and rates of political competition reflected in Polity and Vanhanen dataset, and more subjective assessments of political institutional fabric espoused by the V-DEM by extracting the maximum common variance from each indicator and, thus, partially omitting the noisy component from the common factor. Table 1 summarizes the observable indicators used to extract the common factors of political development in greater detail. In addition, Table 2 reports the spatial and temporal geographic coverage in our sample in greater detail.

Table 1 Measuring distinctive dimensions of political institutions

Table 2 Regional Breakdown of the Sample and the Number of Countries in Factor Analysis

Table 3 reports the estimated principal components of political development based on the pooled analysis of subjective and objective observable indicators used in our analysis. For the sake of internal consistency, we only consider the components with an eigenvalue above 1. In this respect, based on 21,415 country-year paired observations, two principal components are identified from the full set of observable components. The first component explains around 41 percent of the composite variance in the underlying indicator and is characterized by a high eigenvalue above four. The second principal component has an eigenvalue score of slightly below four and explain around 40 percent of additional variation among the remaining indicators. Taken together, both components explain around 81 percent of the overall variation in the political development both across and within countries. Figure 1 depicts the full set of eigenvalues for the whole pooled sample and clearly puts forth that only the first two principal components exhibit eigenvalues above one, indicating an almost natural break between high and low eigenvalues after the second identified principal component.

Table 3 Summary of the principal components of political development

Fig. 1

Screenplot of eigenvalues of the full covariance matrix

Table 4 reports the performed rotation of the loading matrix using Kaiser (1974) varimax criteria that maximizes the variance of the squared loadings within the factors.³ The evidence suggests that the first principal component loads strongly positively on the competitiveness of executive recruitment, openness of executive recruitment, executive constraints, competitiveness of the participation and somewhat less strongly on the regulation of political participation. From the substantive perspective, our interpretation is that the first component synthesizes the formal characteristics of checks and balances given that higher factor loadings are perceptible on the de jure distribution of political power instead of the factual outcomes. By contrast, the second component loads strongly positively on rates of political competition and political participation, electoral democracy index and liberal democracy index as well as somewhat less powerfully on the extent of executive constraints. This implies that the second component synthesizes the overall strength of the civil society and non-elites in controlling and monitoring the political elites and powerholders. Hence, the second component more closely captures the de facto distribution of political power and the ability of non-elites to engage in various forms of collective action and contest against the existing powerholders. Figure 2 plots the rotated factor loadings against each component in greater detail.

Table 4 Rotated components of the loading matrix

Fig. 2

Rotated components loading plot

In Table 5, the descriptive statistics is presented for the latent factors of de jure and de facto political institutions per benchmark year. Panel A exhibits the key parameters for the underlying formal checks and balances series whereas Panel B displays the parametric features of strength of civil society and non-elites’ time series. A lower score on each index corresponds to the more autocratic and exclusionary political regimes whereas a higher score indicates greater extent of pluralism and freedom of participation across the regimes. An overview of the descriptive evidence suggests a notable improvement across both latent variables over time. The mean score on de jure democracy measure tends to increase since the year 1810 onwards, and crosses the zero threshold by 1920 regardless of the sample selection which potentially indicates the prevailing density of slightly more democratic political regime. At the 75th percentile, mean score on formal checks and balances exceeded the zero cutoff already by the year 1900 whilst the score at the 25th percentile of the world distribution has remained below zero throughout the sample period. The mean value of the reconstructed strength of civil society increased proportionately from the initial year onwards without a reversal despite the entry of Eastern European, Central Asian, and Sub-Saharan African countries into the sample upon the data availability before the Third Wave of transition to democracy (Huntington, 1991, 1993).The descriptive statistics clearly suggests the rise of mean score on both principal components from 1810 onwards both at 25th and 75th percentile, uncovering the tendency towards more vibrant and pluralistic formal checks and balances at all layers of political development sweeping across democracies and non-democracies alike. In both cases, the effect of new entrants’ selection into the sample upon the availability of the data can be seen from the rising score range and higher standard deviation over time. Figure 3 plots the mean scores of each component for the full-sample period from 1810 onwards alongside the 95 percent confidence intervals for each year in the period.

Table 5 Descriptive Statistics

Fig. 3

Charting the paths of political development, 1810–2018

Do the principal components of political development exhibit any anomalies in the cross-country distribution over time? To gauge the normality assumption behind the distribution of the scores on first and second component, in Table 6, the normality assumption for the world distribution of formal checks and balances and strength of non-elites is tested using the distribution tests by Kolmogorov (1933), Smirnov (1933), Shapiro and Wilk (1965), Jarque and Bera (1987), and more recent tests by Henze and Zirkler (1990) and Doornik and Hansen (2008). The null hypothesis in each test is that the observed distribution of both principal components follows the Gaussian empirical distribution function. In Panel A, the normality assumption is tested for the world distribution of formal checks and balances whereas in Panel B, the assumption is tested for the distribution of strength of civil society. In both cases, the normality assumption is tested in the base period and the intertemporal sub-samples. The evidence clearly suggests an outright rejection of the null at 1% significance level which tends to obliviate the concerns behind potentially abnormal or anomalous distribution of both variables that could pose inference-related issues.

Table 6 Testing the normality assumption behind the global distribution of democracy, 1810–2018

In Fig. 4, principal components scores are presented for six selected countries for the period, 1810–2018. The evidence clearly highlights several contrasting patterns in the paths of long-term political development. For Denmark and Australia, the score on the strength of non-elites and civil tends to increase substantially over time with short-lived exception in the WW2 period which nevertheless suggests persistently and consistently improved ability to engage in various forms of collective action beyond what is formally specified by the constitutions, laws and electoral systems. A different pattern is perceptible in China where both series exhibit a more stationary dynamics reinforced by the low frequency of changes over time. It should be noted that the component capturing non-elites collective-action capacity has consistently negative score throughout the sample period, indicating the prevalence of more exclusionary and extractive political institutions over time. By contrast, Türkiye’s pattern of political development tends to be persistently less stable where years of steady convergence towards more inclusive and vertically accountable formal checks and balances have been replaced by a series of military coups and civil strife which gives credence to the abrupt fluctuations in both scores. In Russia, the pattern of political development appears to be somewhat similar to China with a notable exception of large-scale reversal of the non-elites’ collective-action component after the ascend of Putin to power in early 2000s.

Fig. 4

Comparative political development in selected countries, 1810–2018

In Table 7, sampling adequacy and relevance of the factor variables in explaining the cross-country differences in both components of political development are assessed in more detail. Several tests are performed to gauge the validity of the reconstructed factor variables. Panel A displays tests of sampling adequacy using the procedure proposed by Kaiser (1974). The test statistics takes the value between zero and one where small values imply that the components have too little cross-variation in common to construct latent de jure and de facto variables using the factor analysis. Our tests demonstrate persistently high-test statistics in the range between 0.80 and 0.87 which suggest the relevance of principal component variables and also implies that the original summary indices indeed do not reflect the maximum variation in each underlying component. As a cross-check, the sphericity test (Bartlett, 1937) is performed for the baseline sample and each subsequent sub-sample under the null hypothesis that the underlying components are not inter-correlated. The null is strongly rejected in the base period and each subsequent sub-period. In Panel B, the country-specific adequacy assessment is presented for the latent factor variables and to reflect the contrasts and similarities in the underlying series. Both latent indices exhibit a strong and persistent similarity since the null on zero correlation is rejected in each sample at 1%. Additional adequacy measures for factor analytical approach to measuring democracy in the long run are computed such as Cronbach’s (1951) alpha and canonical correlation between the formal checks and balances and collective-action capacity measures. Most importantly, we perform the variance equality test between the original summary indices and the latent variables for nine countries from the base sample under the null hypothesis that the variance of the latent principal component variable does not systematically differ from the variance of the original variable. If the variance of the summary index and latent factor is similar, the null is ought not to be rejected which could cast doubt on the use of factor analytical approach towards measuring political development. The evidence advocates the uniform rejection of the null for both measures across all respective countries which confirms the feasibility and relevance of factor analysis in extracting two distinctive measures of political development from the underlying components in the long-term perspective.

Table 7 Assessing the adequacy and relevance of the latent variables of political development

Table 8 assesses the correlation between the principal components of political development and the existing measures and indices of democracy previously established in the literature. The evidence indeed invariably highlights a strong similarity between the principal components and the established indicators of democracy. For instance, our series on formal checks and balances tends to correlate somewhat modestly with the original and revised Polity index as well as with the Unified Democracy Scores (DS) whilst it correlates negatively with the Freedom House scores. In all respective correlations, the null hypothesis can be easily rejected (i.e., p-value = 0.000). Moreover, the component on collective-action capacity tends to correlate somewhat strongly with the Varieties of Democracy summary indices (i.e., p-value = 0.000) as well as with the Freedom House and UDS scores (i.e., p-value), which suggests that our principal components do not uncover any controversial or anomalous finding that would be at odds with the existing literature whilst still improving the aggregation procedure and imparting no fixed end-points on the scale of the component. It should be noted that both components exhibit a statistically significant and relatively strong correlation with the World Bank’s rule of law index in the range between + 0.29 (i.e., formal checks and balances) and + 0.71 (i.e., collective-action capacity) and yield a very low p-value (i.e., value = 0.000).

Table 8 The Correlation Matrix between two principal components of political development and established indicators of democracy

4 Results

4.1 World Distribution of Checks and Balances and Collective Action Capacity

Our results highlight substantial differences in the paths and distribution of formal checks and balances and the strength of civil society both cross-sectionally and over time. Figure 5 reports the world distribution of formal checks and balances for the full period 1810–2018 in designated benchmark years. The vertical line denotes the zero threshold that generally delineates the level of development between more inclusive and open-access (i.e., positive values) and more extractive and exclusionary gradient (i.e., negative values). The evidence readily suggests a marked disparity in the distribution of formal checks and balances from the beginning of nineteenth century. At the beginning of the nineteenth century, the shape of the distribution appears to be substantially left-tailed indicating the preponderance of autocratic regimes characterized by either very weak or largely non-existing checks and balances on the executive powerholders. Until WWI, the form of the distribution tends to evolve into a more symmetric shape which highlights the increasing tendency of a more vibrant democratic development during the first wave of democratization (Huntington, 1991). After the victory of allied forces in WW2, the distribution of formal checks and balances drifts towards a relatively less benign and considerably stronger right tail which predominantly reflects a bold pattern of democratization in the countries defeated in the war such as Germany, Austria, Italy, and Japan. To no surprise, the rise of authoritarian regimes in these by late 1920s and early 1930s was pivotal in the end of the first wave of democratization.

Fig. 5

World distribution of formal checks and balances, 1810–2018

By 1970, the number of entrant countries in the sample with continuous principal component series tends to stabilize which implies that the comparison of the density curves from 1970 onwards portends a relatively clear insight into the dynamics of formal checks and balances over time. In this respect, the density of formal checks and balances inclines to the right, indicating a marked increase in the number of nascent democracies during the second wave of democratization that haunted Latin America (Argentina, Brazil, Peru, etc.), Southern Europe (i.e., Greece, Portugal, and Spain) and parts of the Asia–Pacific region (i.e., Philippines, South Korea, and Taiwan). By 2000, the density curve representing the spread of formal checks and balances is prone to further shift towards the right-concentrated mass, demonstrating the onset of large-scale democratization in eastern Europe and parts of Central Asia after the demise of the Soviet Union. It should be noted that the density curve of formal checks and balances remains relatively stable up to the end-of-sample period (i.e., 2018) although the density right of the zero thresholds appears to diminish somewhat, which indicates the increased prevalence of institutional reversals and democratic backsliding that swept a handful of countries such as Türkiye, Russia, Nicaragua, Venezuela among several others.

A similar pattern can be readily observed in Fig. 6 where the distribution of the institutionalized strength of civil society and non-elites is depicted. Based on the comparison of density curves, the evidence invariably suggests a marked and prolonged shift of the mass to the right over time, reflecting both consistent and persistent democratization through the gradual expansion of suffrage and other forms of expanding access to collective action for the non-elites. By early nineteenth century, only the United States appears to be characterized by a positive score on the strength of non-elites whilst the rest of the world tends to be marked as either more benign and exclusionary (i.e., United Kingdom) or highly exclusionary (i.e., Yemen, Russia, and Ethiopia). By 1870, the number of front-running countries in the expansion of access to collective action increases considerably and includes Switzerland, United Kingdom, Australia, New Zealand, and the Netherlands whilst the most restricted access to any form of collective action for non-elites can be found in Eastern Europe (i.e., Poland and Russia), Africa, and the Middle East.

Fig. 6

World distribution of the strength of non-elites, 1810–2018

The addition of large number of countries with non-missing observations in the base sample by 1900 portrays a heavily left-skewed density curve which may not be surprising given that most of the added countries originate in Sub-Saharan Africa and exhibit very weak constraints on the executive and annihilated access to collective action, which tends to shift the mode of the distribution quite heavily to the left. Prior to the WWII, the disproportionate mass can be identified left of the zero threshold, pinpointing the widespread prevalence of exclusionary practices and institutions that inhibited the access to the collective action for the large cross-section of society. Afterwards, the mode of the distribution shifts consistently to the right accompanied by a pronounced switch in the 1990 during the large-scale political liberalization that swept the countries undergoing the third wave of democratization in Latin America, Southern Europe, Eastern Europe, and Asia–Pacific region. By 2010, the distribution of the institutionalized strength of civil society tends to stabilize and fails to alter the mass by the end-of-sample period indicating both the political turmoil ignited by the Arab Spring and the erosion of democratic accountability amid the rise of autocratization (Lührmann & Lindberg, 2019).

4.2 Charting the Paths of Political Development, 1810–2018

Are the estimated paths of political development in conformity with the existing measures commonly used in cross-country comparisons? Fig. 7 charts the trajectories of both principal components for the United Kingdom and the Netherlands which serve as well-known and commonly discussed cases in the political science literature (North & Weingast, 1989; Van Bavel, 2020; Van Zanden & Prak, 2006; Weingast, 1993). The evidence unveils both substantial similarities and a number of notable differences between our series and the existing measures of formal political regimes (Marshall & Gurr, 2020) and measures of electoral and liberal democracy (Coppedge et al., ). Our series uncover evidence of gradually reinforced improvement in the formal checks and balances as well as a growing strength of the non-elites’ capacity for collective action towards heightened and more rigorous elite accountability prior to the beginning of WWI. Our series of formal checks and balances indicates the first major turning point in the political development of the United Kingdom in 1832, which coincides with the First Reform Act, which redistributed the seats in the lower house of Parliament and expanded suffrage through the electoral reform. The second most pivotal juncture in the series arises in 1928 when the Equal Franchise Act expanded suffrage to women aged 21 and above, indicating a massive expansion and movement towards a universal suffrage. A continuity of stable, pluralistic and relatively inclusive political development persists throughout the twentieth century remaining both intact and unhinged. Domestic political instability by the end of 2016 coupled with the United Kingdom’s exit from the European Union yields a noteworthy downward spiral of formal checks and balances which is consistent with Marshall and Gurr (2020) assessment, although the decline in the strength of formal checks and balances is somewhat stronger in our series. By contrast, it should be noted that no such decline is perceptible in the series on the capacity of non-elites for various forms of engagement in collective action which implies that the notion of democratic backsliding discuss earlier should be taken with some caution (Waldner & Lust 2018, Grillo & Prato, 2023).

Fig. 7

Charting the trajectories of political development in United Kingdom and the Netherlands, 1810–2018

Similar pre-WWI trajectory of political development unfolds in the series on formal checks and balances and strength of non-elites and civil society for the Netherlands. The first major juncture appears in 1848 when the constitutional reform (i.e., Grondwetsherziening van 1848) promulgated the institutional foundations of the parliamentary democracy, granting the House of Representatives more influence and establishing direct elections. The second most salient juncture is posited in 1910s when a less restrictive form of suffrage is extended to the broader cross-section of society whilst the universal suffrage is achieved by 1919 when the House of Representatives voted in favour of women’s suffrage by a large majority, setting the Netherlands on the trajectory of political development comparable with the United Kingdom. During the occupation by nazi Germany, both series plummeted, and tend to rebound immediately after Germany’s defeat in WWII and the subsequent restoration of parliamentary democracy. Throughout the post-WWII period both the trajectory of formal checks and balances as well as the strength of civil society and non-elites remain stable with no evidence of reversal, backsliding or decline. Pointwise, the correlation coefficient between the formal checks and balances and the Polity2 score is + 0.84 (i.e., p-value = 0.000) for the Netherlands and + 0.79 (p-value = 0.000) for the United Kingdom. The corresponding correlation coefficient between strength of civil society and Coppedge et al., (2021) liberal democracy series from V-DEM dataset is + 0.98 (p-value = 0.000) for both countries, respectively.

Figure 8 presents the trajectories of political development for Spain and Argentina for the period 1810–2018. Contrary to the United Kingdom and the Netherlands, the contours of political development in Spain and Argentina endured substantially greater turmoil and institutional instability in the twentieth century.⁴ Yet, it should be noted that embryonic and incessant instability is not per se the defining feature of the political development of both countries from the beginning our period of investigation. Against the setback of the legacy of Napoleonic invasions that left Spanish society weak and unstable, the first major juncture arises in 1870 when the First Spanish Republic is declared by the republican and radical forces. The prevented formation and consolidation of the republican form government by a military coup in 1874 laid the foundations for the subsequent restoration of the monarchy in subsequent years. This particular juncture is clearly indicated by an abrupt decline of the capacity of the civil society to engage in collective action in the subsequent years whilst the erosion of formal checks and balances is precipitated in early 1890s when the intellectual movement to demanding liberal changes was suppressed and crackdown after several progressive professors at the Central University of Madrid promoted the ideas of German philosopher Karl Christian Friedrich Krause. The elite suppression of liberal movement largely reinforced traditional values that gained momentum during the rise of Falangists (Jensen, 1999). Afterwards, Spain endured a vicious cycle of political development characterized by a weak civil society and unconstrainted executive powerholding. The institutional progress during a short-lived establishment of the republican form of government in the 1930s commenced with the introduction of universal suffrage. However, the progress came to an abrupt end after a series of class conflicts between the traditionalists, supported by the Catholic Church, and the progressive-left nexus escalated into the military coup d’etat against the republican government, and precipitated three years of civil war (La Parra-Pérez, 2020). After the rebel victory and establishment of dictatorship under Franco, we observe a nearly complete demise of both the formal checks and balances as well as the rampant erosion of the capacity of non-elites and civil society to engage in collective action. In late 1970s, the political liberalization and subsequent institutional transition towards democracy is associated with a permanent improvement of both formal checks and balances and non-elites’ collective-action capacity.

Fig. 8

Charting the trajectories of political development in United Kingdom and the Netherlands, 1810–2018

By contrast, Argentina’s trajectory of political development exhibits a somewhat greater tendency of stability after the new constitution was promulgated in 1853. Unlike in Spain, the trajectory of formal checks and balances tends to be relatively stable under a series of liberal governments although the political order exhibited few open-access characteristics (North et. al. 2009). The first major turning point in the path of formal checks and balances arose in 1912 when Sáenz Peña Law established universal, secret and compulsory male suffrage through the creation of electoral lists. Our series reinforce the view that the checks and balances as well as the non-elites’ collective action capacity precipitously declined during the 1930 coup d’etat and subsequent rise of Péron to power in 1946 (Alston & Gallo, 2010). The military coup d’etat in 1955 that removed Péron from public office failed to curb the dynamics of embedded institutional turmoil. The series of unstable and fragile govenments by the government administrations led by Union Civica Radical (UCR) was interrupted by another military coup d’etat in 1966, subversive instability and class conflict that eventually culminated into the military dictatorship between 1976 and 1983 (Spruk, 2019). After the political liberalization and institutional transition towards democracy after 1983, our series indicate a stable development of checks and balances and a non-elites capacity to engage in various forms of collective action although Argentina tends to lag behind mature democracies such as the Netherlands, Switzerland and Germany.

4.3 Revisiting Three Waves of Democratization

Major surges of democracy have long been a subject of vibrant debate among political scientists and economics that has been dubbed “waves of democratization” originally proposed by Huntington (1991). The general thrust of the argument is that waves of democratization embolden sudden shifts if the distribution of political power, further stimulating political liberalization and domestic institutional reforms. In the original formulation, Huntington describes three waves: (i) the slow wave of the nineteenth century, (ii) the second wave after the allied victory in WWII, and (iii) third wave beginning in mid-1970s in southern Europe and gradually expanding to Latin America, Asia–Pacific and Central and Eastern Europe. It should be noted that the definition of waves is susceptible to the analysis and assumptions (Gunitsky, 2018).⁵

The question that remains is whether the latent series capturing formal checks and balances as well as non-elites collective action capacity confirm or negate previously established trends of democratization over the past two centuries. The empirical evidence from the reconstructed principal components of political development reinforces the notion of different waves of democratization. Figure 9 charts the trajectories of formal checks and balances and collective action capacity of civil society during Huntington (1991) first wave of democratization encompassing political liberalization through early suffrage extension beyond property-owning class in mid-nineteenth century in Western Europe, Western Offshoots and Argentina. The evidence confirms a general upward trend in the latent series on formal checks and balances. For instance, United Kingdom and Australia appear to have the strongest formal checks and balances by late 1830s and stand out against considerably weaker checks and balances in Argentina and France. Prior to the end-point year in 1918, the level of former checks and balances exhibits some sign of convergence despite the widespread heterogeneity between Canada and Australia at the upper tail and Italy at the lower end. Similar trends can be observed in the evolution of the non-elites’ collective action capacity. By early nineteenth century, United Kingdom, and, to a lesser extent, Australia tend to have the strongest capacity of civil society and non-elites to challenge the existing powerholders by institutionalized modes. Prior to the end of the Victorian era, the first wave countries fall apart between United Kingdom, Australia, Canada and France on one hand where a rapid improvement of the capacity is perceptible, and Italy and Argentina on the other, as they tend to lag behind Western Europe considerably. Prior to the end of WWI, both countries tend to move closer to the political development frontier.

Fig. 9

Charting the political development during the first wave of democratization

Furthermore, Fig. 10 uncovers the trajectories of political development for two countries that underwent the institutional reversal from early democratization into full-fledged authoritarian regime in the 1930s, using the cases of Germany and Austria. The gradual development of the formal checks and balances uncovers a stepwise transition from the authoritarian constitution under the Bismarck chancellorship (1862–1890) to the more open-access political order through institutional reforms such as three-class franchise until the 1890s when Social Democratic Party was founded, imparting a significant pressure on the elites for political reforms towards a broader representation in the parliament. A somewhat similar trend is apparent in the development of the collective-action capacity amid a setback during the conservative authoritarian rule under Wilhelm II. Against this backdrop, Germany’s post-unification political development, whilst converging with Western Europe in terms of the formal checks and balances, significantly lagged behind in terms of the capacity of non-elites to engage in various forms of collective action. Similar trajectory of political development is charted for the former Austrian Empire where a more extractive regime of checks and balances is indicated prior to WWI consistent with both Freudenberger (1967, 2003) and Acemoglu and Robinson (2006b) where abundant evidence and discussion on elite-blocking institutional and technological development of the Austrian empire prior to the Hapsburg empire collapse is provided (Gerschenkron, 1977). It should be stressed that the collective action capacity for the civil society and non-elites steadily improved and gradually converged with both Germany and, to a lesser extent, Italy, while lagging behind the political frontier. It should be noted that the series of formal checks and balances for Germany prior to WWI indicates a levelling with the countries undergoing the first wave of democratization in Fig. 9. However, Germany’s political development in pre-WWI period distinguishes itself from North-Western Europe and Western Offshoots by a relatively benign and weak civil society lacking both the capacity and access to challenge the entrenched elites’ power.

Fig. 10

Charting the political development trajectories of German and Austrian Empire, 1810–1918

After the defeat in the WWI, the trajectories of political development in Austria and Germany improved dramatically after externally-imposed institutional reforms by the Allied forces established multi-party democracy. Figure 11 portrays the trajectories of both principal components in Austria, Germany and Japan before and after their defeat in WWII. The trajectory of reconstructed principal components emphasizes a strong degree of institutional stability in the development of formal checks and balances. None of the three countries defeated in WWII is characterized by the institutional reversal away from checks and balances towards their intrinsic erosion which indicates a somewhat permanent effect of the large-scale political liberalization. A similar insight unfolds in the comparison of the intertemporal evolution of the strength of the civil society. Whilst the pre-war strength of non-elites in the three countries was severely undermined and decimated by the authoritarian regimes in place, post-war development reinforces a pattern of a vibrant and active civil society underpinned by widespread and stable access to collective action for the non-elites. The series also appears to be relative stable and similar, indicating little heterogeneity in the trajectory of political development after the end of WWII.

Fig. 11

Trajectories of political development after the allied victory during WWII in selected countries, 1918–2018

Furthermore, Fig. 12 reviews the early-moving countries in the third wave of democratization in the 1970s and 1980s that haunted both southern Europe, Yugoslavia, and Asia–Pacific region as well as the former Soviet orbit at the later stage, and that is a subject of vibrant discussion (Haggard & Kaufman, 2016). More specifically, the figure plots the trajectories of formal checks and balances and strength of civil society before after the year of democratization (Doorenspleet, 2000) that is indicated by the vertical line. The evidence largely confirms an omnipresent pattern of rigorous democratization reinforced by the permanent improvement in the rigor of formal checks and balances as well as by the noteworthy and substantial expansion of the capacity of the civil society to engage in the variety of collective action forms. The institutional development of the polities after the democratization appears to be somewhat uneven. In southern Europe, the political liberalization commencing with the fall of the military rule coincides with an immediate and permanent improvement of the checks and balances promulgated by the new constitutions. In Asia–Pacific sample, the strengthening of formal checks and balances is particularly slower and more cumbersome with the exception of Taiwan. By contrast, the development of the collective-action capacity for the non-elites appears to be quite homogenous across all affected countries with few signs of either decadence or reversal up to the present day. Thus, our series suggests that the third wave of democratization promulgate two distinguished direction of political development. First, the political liberalization in southern Europe in late 1970s and early 1980s promulgated both strong formal checks and balances on the executive powerholders accompanied by a more vibrant and active civil society underpinned by the widespread improvement in the capacity for collective action. And second, the democratization in Asia–Pacific region appears to revolve primarily around stronger and more resilient civil society but in the presence of relatively weaker formal checks on the abuse of the executive power. Nonetheless, the trajectories of political development either in terms of formal checks and balances or non-elites’ capacity for elite replacement tend to be congruent and indicate an episode of irreversible political liberalization up to the present day.

Fig. 12

Political development trajectories during the third wave of democratization in affected countries, 1950–2018

The second part of the third wave of democratization is exhibited in Fig. 13 which represents the trajectories of both principal components of political development for a group of central and eastern European countries. A peculiar characteristic of these countries is that prior to the democratization, they were part of the eastern bloc and under the control of the Soviet Union prior to its dissolution in 1990. Our series of political development for the central and eastern European countries highlight a pattern of both rapid and large-scale democratization that emphasizes a gradual but persistent heightening of the formal constraints on the powerholders. In some affected countries such as Hungary, Poland and Slovakia, the expansion of checks and balances appears to have been rapid and immediate. Yet, the newly established democracies in central and eastern Europe clearly lag behind the political frontier and tend to have considerably weaker formal constraints on the executive. Some countries in the region also tend to experience well-known episode of stalled institutional reforms of checks and balances. This is best indicated in the series of formal checks and balances in Bulgaria and Czech Republic. The latter is characterized by stalled progress towards strengthened checks on the executive branch since 2000 whilst the latter embarks on the perilous path of the electoral reform in the period 1998–2000 which introduced many biases against smaller political parties in terms of the proportionality of the representation, shifting the electoral system substantially towards the majoritarian direction. In turn, our series on formal checks and balances for Czech Republic exhibits a marked and one-off deterioration in 2002.

Fig. 13

Political development trajectories in Central and Eastern Europe, 1970–2015

The development of the civil society and the capacity of non-elites for the engagement in forms of collective action follows a slightly more distinctive contour and is marked by widespread heterogeneity. In the 1990s, the newly established democracies in Central and Eastern Europe emboldened a reasonably strong civil society indicated by an almost discontinuous jump in the series on the component capturing non-elites access to collective action. The initial progress in the collective action’s access trajectory appears to be broad and particularly widespread in countries such as Poland and Czech Republic which appear to reach some parity with the established collective action capacity in Western and Southern Europe. By contrast, some countries in the region, especially Hungary, witnessed a substantial deterioration in the strength of civil society during the rise of political parties such as Fidesz, that openly advocated the alternative model to the Western liberal democracy. For Hungary, although the erosion of formal checks and balances does not seem to be perceptible, the deterioration of the civil society’s strength and non-elites’ capacity for collective action tends to be widespread and coincides with the rise of Órban to power. A similar decline but to a lesser degree is perceptible in Poland, Bulgaria, and Slovakia.

The rise of reform movements in the Middle East that later turned into armed rebellions represents another important turning point in the paths of political development that merits some discussion. Figure 14 presents the trajectories of formal checks and balances in some countries covered in our sample that were affected by the Arab Spring. The vertical line in each graph denote the year in which the protests and rebellion began. The evidence uncovers substantial heterogeneity in the trajectories of political development after the Arab Spring. The disparity in the post-Arab spring political development can be decomposed into two insights. First, the strength of formal checks and balances does not appear to improve inasmuch as during the previous waves of democratization. Neither of the countries in our sample exhibits a marked improvement in the strength of formal checks and balances except Tunisia where checks and balances tend to heighten after a comprehensive constitutional reform in 2014. Elsewhere in the Middle East, our series fails to indicate designated improvements in the rigor of checks and balances after the Arab Spring. In most cases, such as Libya, Syria, Egypt and to a lesser extent Yemen, the political turmoil in the post-Arab Spring period further weakened and eroded the formal checks and balances on the executive power. Civil war in Syria, the rise of Muslim Brotherhood in Egypt and the subsequent coup d’etat in 2013, armed insurgency in Yemen and civil war in Libya that began 2014 comprise a few salient domestic institutional changes and armed conflicts that can be associated with repeated failures to strengthen formal checks and balances. And second, despite some incremental progress in the development of the non-elites’ capacity for collective action, our series indicates Tunisia as the evidence of relatively successful political liberalization whilst the non-elites’ capacity improvement elsewhere in our Middle East sample is either temporary and short-lived (i.e., Egypt), non-existent (i.e., Jordan and Syria), or deteriorates further (i.e., Yemen). The failure to develop a vibrant civil society capable of the engagement in the collective action process has been linked to the historical legacy of oppressive political institutions and fragmentation (Kuran, 2023; Malik & Awadallah, 2013).

Fig. 14

Political development trajectories in the Middle East before and after the Arab Spring, 1970–2018

And lastly, an important defining characteristic of the political development in the past decades haunted the retreat of democracy and the subsequent autocratization of politics which has been dubbed democratic backsliding (Mechkova et al., 2017; Waldner & Lust, 2018), democratic erosion (Kneuer, 2021) and fourth wave of autocratization (Lührmann & Lindberg, 2019). A handful of case studies where the democratic erosion unfolded has been discussed such as Russia (Petrov et al., 2014), Türkiye (Bayulgen et al., 2018), Hungary (Enyedi, 2018, Haglund et al., 2022), Nicaragua (i Puig & Serra 2020), and several others. The general thrust of these studies emphasizes emboldened autocratization of political institutions through populist-based erosion of checks and balances, undermining of judicial institutions and government-orchestrated assaults on civil society. The question that arises immediately is whether the recent wave of autocratization that began to unfold gradually since in early 2010s can be detected in our reconstructed series on checks and balances and non-elites collective-action capacity?

Figure 15 charts the trajectories of formal checks and balances and non-elites’ collective action capacity for the selected countries that underwent the third wave of autocratization. The vertical line indicates the year in which populist powerholders were elected and assumed office. The evidence largely confirms substantial and heightened reversal of the political development. Several insights can be drawn from our series concerning the fourth wave of autocratization. First, the wave of autocratization primarily operates through the erosion of the collective-action capacity as the respective principal component deteriorates immediately after the autocratizing politicians assume power. For instance, the strength of non-elites collective-action capacity deteriorates immediately after the rise of Viktor Órban, Law and Justice Party and Vladimir Putin to power in Hungary, Poland, and Russia. In some of these cases, the autocratization does not materialize through the institutionalized erosion of formal checks and balances as the evidence from these countries indicates, as they tend to preserve the formal checks and balances in place, and deploy other forms of institutional technology (Scartascini & Tommasi, 2012) to implement their policies and power-consolidating strategies, primarily through either formal or informal assaults on the civil society that has been subjective to ongoing debate (Esen & Gumuscu, 2017; Guriev & Treisman, 2019). Second, the rise of autocratization outside Europe is oftentimes accompanied by both dismantling of formal checks and balances and the decaying civil society which, in turn, annihilates the non-elites access to collective action. The evidence in support of such variety of autocratization is found in Türkiye and Venezuela where both components decline after the consolidation of Erdogan’s power grip in 2012 and the rise of Chavista movement in 1998 that facilitated a radical institutional overhaul towards the near disappearance of both formalized checks and balances and civil society’s ability to engage in the collective action against the powerholders. The distinction between the former and latter type of autocratization has been recently defined as the distinction between spin-dictatorship and fear-dictatorship (Treisman & Guriev, 2023) where the former are characterized as media-savvy redesign of authoritarian rule whilst the latter tend to deploy violence, fear and ideology to cement the authoritarian rule (Egorov & Sonin, 2011) and implement economic policies receiving significant support from the majority of the population that ultimately hurt their interests (Acemoglu et al., 2013; Boix & Svolik, 2013). And third, the democratic backsliding in the United States after the election of Donald Trump is a subject of rigorous debate and discussion (Grumbach, 2023; Kaufman & Haggard, 2019).

Fig. 15

Democratic reversals during the third wave of autocratization in selected countries, 1970–2018

And third, the democratic backsliding in the United States after the election of Donald Trump in 2015 is a subject of rigorous debate and discussion (Lieberman et al., 2019; Kaufman & Haggard, 2019). The estimated principal components of political institutions for the United States indicate two important insights to be drawn from post-2015 series. First, unlike elsewhere where populist came to power, Trump’s presidency buttressed substantial corrosion of formal checks and balances through toppling the boundaries of civil disclosure, rule by intimidation, a complete obliteration of the balance of power, sowing distrust in the electoral system, and lastly, inspiring the resurrection against the government during the Capitol Hill attack on January 6 among many other examples. Contrary to the backsliding episodes elsewhere, the United States still preserved the relatively strong and resilient civil society that was instrumental in the ultimate end of Trump’s presidency in 2021. Although the principal component on the non-elite capacity indicates an erosion after the ascend of Trump to power, the decline does not seem to be historically unprecedented in the lieu of prior temporal shifts in the political development of the United States.

Since an exhaustive discussion of the diverging paths of development is beyond the scope of this article, the contrasting paths of political development in the long-term perspective and data-driven identification of country-specific major turning points are discussed in greater detail in the supplementary appendix.

4.4 Estimating the Ideal Points of Political Development

One of the caveats behind the internal validity of our principal components hinges on the aggregate uncertainty and reliability of our estimates. Although the constructed principal components tend to exhibit a rather strong correlation with the existing indices in the literature on political development, the validity can be partly compromised if both series were contaminated by substantial noise. Therefore, the construction of the confidence intervals around a plausible and interpretable significance threshold is essential in lending credence to the internal validity of our estimates. Since the construction of confidence intervals on the eigenvector-determined principal components is subject to severe intrinsic limitations and is not possible without asymptotic standard errors or bootstrapping-based empirical strategies (Timmerman et al., 2007), our approach follows the comparative political economy literature (Pemstein et al., 2010) in which political institutions cannot be fundamentally observed are considered as latent traits for which ideal points can be constructed (Martin & Quinn, 2002, Imai et al. 2016, Spruk & Kovac, 2019) through the application of Monte Carlo Markov Chain numerical approximation and sampling algorithm. This approach proceeds in two steps. In the first step, a Bayesian measurement model of possible ideal points is fitted to obtain the full distribution of possible scores of formal checks and balances and collective-action capacity through randomly sequenced MCMC algorithm based on the target probability distribution either with or without prior assumptions on country-specific mean and standard deviation parameters. In the second step, confidence intervals are easily constructed from the entire possible distribution of latent scores for each country-year pair. We proceed in two steps. In the first step, we estimate ideal points of both principal components assuming non-zero drift. This implies that for each $t=\text{1,2},\dots T$ we fit a Bayesian item response model through the application of Metropolis–Hastings algorithm with a flat prior distribution of parameters to obtain repeated cross-section of the ideal points where country-year-specific confidence intervals are constructed. In the second step, we relax the institutional drift assumption, and estimate a dynamic Bayesian item response model based on the Gaussian distribution, and flat prior assumption for each country $i=\text{1,2},\dots N$ over the full period of non-missing observations. This yields dynamic country-specific trajectories of political development for both components as well as the respective 95% confidence intervals. To allow for the political development trajectories to trend smoothly over time without imposing prior assumptions on the direction of development, we cap country-level sample size at 1,000,000 observations which yields roughly 34 billion iterations of the baseline Bayesian item response model in a prior-free learning environment. Furthermore, to separate the structural and cyclical component of the political development, we compose a high-frequency filtering parameter (Ravn & Uhlig, 2002) to separate the long-run trend of political development trajectory from its cyclical component.

The key advantage of MCMC estimated ideal points stems from a comparable scale denoting the degree of inclusiveness both across and within countries and facilitates a direct and rather straightforward comparison accompanied by the respective confidence bounds behind year-specific estimates of ideal points. Because the a continuous and time-dependent fit of the Bayesian item response model solves fits a dynamic trajectory of political development and embed drift component into the chained sequence of the sample and eliminate potential noise, dynamic ideal points of political development are relatively more informative on the long-run variance in the trajectories of political development across countries. Although full and exhaustive comparison of the dynamic trajectories of formal checks and balances and non-elites’ collective-action capacity is beyond the scope of this paper,⁶ Fig. 16 reports the dynamic ideal points of formal checks and balances for selected countries. The evidence confirms prior patterns of formal checks and balances’ evolution over time. For instance, countries such as Australia and Sweden exhibit a relatively stable evolution above the zero-threshold from early nineteenth century onwards followed by a marked and permanent shift towards the frontier upon the major respective institutional changes that corresponding to the establishment of the Commonwealth of Australia (i.e., in 1901) and large-scale social reforms by Social Democratic Party in early 1920s and subsequent extension of suffrage to women. Furthermore, dynamic ideal points also unveil non-trivial reversals in the inclusiveness of formal checks and balances such as a severe erosion of checks and balances in Türkiye after Erdogan’s constitutional referendum in 2010 and subsequent complete undermining and assault on the judiciary and crackdown of civil society, and South Africa’s persistent instability and more gradual erosion of the checks and balances from late 1990s onwards.

Fig. 16

Ideal points of formal checks and balances in selected countries, 1810–2018

Furthermore, Fig. 17 reports several noteworthy country-specific trajectories of non-elites’ collective action capacity. Dynamic ideal points largely confirm widespread country-specific differences in the development of collective-action capacity over time and contrasts stable and inclusive episodes of political development in countries such as Denmark and, to a substantially lesser extent, United States, against either stationary and exclusionary episodes and deep reversals such as Türkiye under Erdogan, Russia under Putin’s rule and Cuba under Castro’s regime. Additional examples ought to be exhaustively analysed and assessed in more rigorous data-driven perspective to discuss the sources of large and wide-standing differences in political development over time.

Fig. 17

Ideal points of the strength of civil society and non-elites

5 Conclusion

This paper presents an attempt to measure and quantify the trajectories of formal checks and balances and non-elites collective-action capacity for 167 countries for the period 1810–2018. Our approach overcomes the existing aggregation and scale validity bias that contaminate existing long-run democracy measures by addressing the measurement error in the extant democracy indices. Compared to the earlier attempts by Pemstein et al. (2010) and Trier and Jackman (2008), our approach decomposes the components of democracy into (i) formal and (ii) and outcome-related ones, and focusses on the long-term dimensions of political development rather than post-WW2 perspective for which a substantial number of studies and datasets is available. In this respect, the longitudinal variation in the de jure and de facto characteristics of political systems is exploited using the formal features of the political regimes (Marshall et al., 2013) and objective rates of political competition and participation (Vanhanen, 2000, 2003), as well as the new generation of expert-based assessments of the state of democracy (Coppedge et al., 2021).

Using the latent factor analytical approach, two continuous latent measures of political development are derived which allow for the comparative institutional analysis of political regimes in the long-term perspective to examine the effects on a broad array of socioeconomic and institutional outcomes. The proposed latent measures are used to construct the world distribution two principal components of political development using the density curves which demonstrates some limitations of the summary-based indices indices in measuring the democracy in the long-run stemming from the arbitrary weighting scheme, scale validity issues and aggregation bias. Our latent analytical approach overcomes such bias with the synthetic de jure and de facto measures of democracy which emphasize both the existence and persistence of multiple peaks in the distribution of democracy across countries from the initial year onwards. The evidence from multivariate analysis confirms the relevance of principal component analysis in extracting both aspects of political development from the underlying components into two unobservable synthetic measures and we further demonstrate the reliability of the obtained factor scores.

The patterns from the synthetic measures uncover the persistent and steady democratization. The multiple peaks in the distribution of de jure checks and balances already appear in early nineteenth century and do not disappear when additional countries are selected into the sample. On the other hand, the twin-peak distribution of collective-action capacity already appears upon the start of the nineteenth century and remains stable across multiple decades even after controlling for sample selection effects. The factor scores for de jure and de facto democracy suggests a contrasting pattern between the two. De jure political liberalization increases persistently until 1850 followed by a drop until 1920s. It also improves continuously until 1970s and survives the recent resurgence of autocratization. The pace of collective action-related capacity expansion increases continuously from 1810 until 1920 and exhibits a strong resurgence after the Allied victory in WW2. Concerning the recent debate over measuring democratic declining or backsliding, our findings confirm some reversal but, in line with other scholars in comparative politics (for example, Treisman, 2023), not enough to jeopardize long-term institutional trends.

Using comparative country case studies on the trajectories of political development, our evidence demonstrates substantial differences in the speed and shape of development marked both by convergent similarities and divergent contrasts that amplify considerable over time.⁷

This article adds to the nascent and growing empirical political economy literature a latent analytical approach to precisely measure the shape of political development as a continuous institutional variable rather than a dichotomous variable or a concept too broad to be measured. Our work presents an attempt to extract the most powerful and plausible latent measures from the underlying components in the long-term perspective which allows for the comparative analysis of the long-run political development. Factor analytical approach is adopted to plausibly reweigh and reconstruct existing indices of democracy that allow us to extract the maximum variance from the underlying indicator that is hidden from binary measures or other measures that rely on a single index. Concurrent to Pemstein et al. (2010), this article adopts the posterior distribution approach and extends the temporal coverage to a more long-term perspective that enables us to chart the patterns of political development using latent indices. In the quest to improve the existing estimates, the building block of future research should rely on improving and delivering more detailed measures of the state of democracy related to franchise and electoral competition that would allow more in-depth insights into the differential paths of political development, and identify the major turning points in clear data-driven perspective.