Understanding Illiberal Democracies, Liberal Autocracies, and Everything 2

Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between economic activity, and strive for equality of opportunity. As designed, there are 7 points to the scale, with a “1” being applied to the countries that fits this ideal most effectively (e.g., Argentina 1986-1988; New Zealand) and a “7” being applied to “countries and territories (with) virtually no freedom” (e.g., Haiti 1991-1993; Albania 1972-1989) (McColm et al 1991, 72). Although indicators for everything tend to be criticized, this one in particular has been particularly targeted. Some have been “put off” the measure because it is not clear how the coders derive their rankings. Unlike the Polity measure, there is no clear breakdown of what falls into different categories. Some have suggested that the measure was inappropriately used as an indicator of repression when it seemed to be representing “democraticness.” I have tended to agree with these criticisms in the past, especially since my research has been based in human rights, while trying to assess the impact of democracy on this behavior. For the purposes of this study, however, and the generally positive evaluations that have been given to the measure in the recent past, I feel more confident about its application here. BRINGING TOGETHER THE DIMENSIONS In line with the figure provided earlier (Figure 1), to identify different polity structures I overlay the POLITY and Freedom House measures, using this as a basis to identify different regime types. The biggest problem with this effort arises in designation of “cutoff points”. Whereas other conceptualizations of polity structure are directly comparable to the one identified above (e.g., Bratton and Van de Walle’s [1994, 471] as well as Dahl’s [1971, 7]), in this case I am forced to be a bit more precise by specifically identifying when Illiberal Democracies end and Liberal Democracies begin? Additionally, I must address where Liberal Autocracies begin and Illiberal Autocracies end? For this paper, understanding that the subject of proper assignments will require much more attention that is possible here, I have decided to put forward suggestions that rely upon the information provided within the two codebooks in conjunction with how individuals have actually used the data in the past. 14 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between For the POLITY measure, one could make a case that regimes above a score of “8” are fully democratic (i.e., political rights exist) and those below this score are autocratic (or at least non-democratic). Within the former context, there are constraints on the chief executive, some form of competitive executive recruitment, some openness of the selection process, and some competitiveness in political participation. Within the latter, the various characteristics are more or less opposite (e.g., there are no constraints on the executive and so forth). For the Freedom House measure, the authors are more explicit about where they would place different regimes (i.e., where they would place the “cut off” point). From the coding procedure, category “2” (e.g., Panama 1990-1991) is relatively free and “3, 4 (and) 5 have progressively fewer civil liberties than those in category 2 (i.e., they are partly free)” (McComb et al 1992, 72). Using this information, I establish “3” (e.g., Albania 1992) as the cut off point and argue that those that fall above this category are illiberal (e.g., Hungary 1987-1988 in category “4”) and those that fall below it are liberal.18 This is displayed below in Figure 2. (Insert Figure 2. About Here) EXPLANATORY VARIABLES As discussed above, there are numerous variables that are employed to account for variance in polity structure. As many are discussed elsewhere, my discussion of them is brief. Internal Factors. The Duration of Polity structure (i.e., the number of years that a specific government has been around without significant change in its institutions [changes in the Polity score equal to or greater than 2 points]) is taken from the Gurr (1999) data. Protest is operationalized in line with the indicators developed by Davenport (1995). As designed, several dissident events are drawn from Banks’ Cross-National Time-Series Data Archive (1998): strikes, guerrilla warfare, riots and protest demonstrations, and these are used to construct three measures: 1) cumulative frequency (i.e., the number of dissident events taking place during a particular year), 2) the degree of violence present (i.e., whether or not guerilla warfare or riots 15 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between take place), and 3) the deviance of protest behavior from the cultural norm (i.e., the degree to which dissent within a particular year deviated from the mean of the country’s experience over the full time period under investigation). Banks was also used to capture more elite forms of conflict behavior such as Coups (successful and attempted); large-scale, mass behavior exceeding the parameters of protest in the form of Revolution, and Population (expressed in 100,000’s). External Factors. Following the suggestions of summers and Heston (1988), who identify the problems with the use of GNP and GDP, Economic Development is measured with energy consumption per capita. This measure is also taken from Banks’ (1998). There are four indicators for spatial context; all are drawn from the work of Monty Marshall (2000). The number of Countries on the Border represents the “Number of states with common land border to unit-specific state” (Marshall 2000, 3) and the number of Countries in the Region represents the number of countries within specific geographic locales. 19 Two measures are provided for the magnitude of conflict experienced within the bordering countries (i.e., Bordering Armed Conflict) and the region (i.e., Regional Conflict). Relevant conflicts include: international violence, international war, international independence war, civil violence, civil war, ethnic violence, and ethnic war. Considering such factors as human resources (deaths, injuries, etc.), forced dislocations, and infrastructural damage, these indicators are coded on a scale from “1” to “7” (with “7” being the most severe) for each of the countries involved. This score is then summarily added to identify the total magnitude of conflict within neighbors or the region, respectively.20 METHODOLOGICAL TECHNIQUE To analyze variance in polity configuration, I employ the use of a Multinomial Logistic Regression (e.g., Aldrich and Nelson 1984; Hosmer and Lemeshow 1989). This strategy is utilized for there are four outcomes that are possible for the dependent variable, which are all 16 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between mutually exclusive and unranked. The latter characteristic is most important as (STATA 1999, 399) (i)t is the unordered categorical property of (the dependent variable) that distinguishes the use of (multinomial logit models) from (regression) (which is appropriate for a continuous dependent variable), from (a model) (which is appropriate for ordered categorical data), from logit (which is appropriate for two outcomes and which can therefore be thought of as ordered).21 The model proves to be particular useful in the context of this analysis because it allows us to simultaneously examine conditions that are associated with polity structure as well as factors that will lead to other categories (through the examination of causal impacts across different values of the independent variables). The effort at comparative assessment is facilitated through the selection of the base-category (i.e., the value of the dependent variable that forms the basis of comparison across different values). Since the majority of countries in the world have historically been within the Illiberal Autocratic or Liberal Democratic categories, these were used as the base categories for estimation. Results thus identify the probabilities that different polity structures exist given the existence of particular independent variables. Finally, as the coefficients reported from multinomial logit models are somewhat confusing (frequently sending individuals into discussions of “incidence rate ratios” – which are somewhat difficult to interpret), I use King et al.’s (1999) computer simulation Clarify to present results. This model uses Monte Carlo simulation to convert statistical results into theoretically and practically meaningful values. For this paper, results are computed into probabilities such that I will be able to say, for example, that when variable x takes a value of v events that the probability that the polity structure is going to be “3” (Liberal Autocracy) is z%. 17 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between 3. EXPLORING POLITY CONFIGURATIONS: A FIRST LOOK Perhaps the best way to begin the analysis is with a general overview of polity characteristics. As noted by numerous scholars (most notably Huntington 1991), the specific type of regimes has changed significantly over time. Drawing upon Gurr’s POLITY measures (Figure 3 in the upper left corner, expressed in percentages), we are able to see that since 1810, there have been three waves of democratization (from 1828-1926; 1943-1962; and 1974 to the present) which have each (after periods of retraction) increased the number of democracies (as well as polities) found within the world. For clearer exposition, on the right hand side I have highlighted the third wave. Of course, this only considers the political rights dimension. When one factors in consideration of political liberties, then we are able to gauge the distribution of diverse political configurations discussed earlier (unfortunately, this can only be done for the 1972-1996 time period). (Insert Figure 3. About Here) Viewing the figures on the bottom, one can see that Illiberal Autocracies are still dominant as a percentage of all polities. Most of the political systems, even amidst the current period of global democratization, live in relatively autocratic governments. Liberal democracy (as a regime type) has made significant inroads during this third wave and thus by the 1996 time period, the first two polity configurations begin to converge in their presence within the global system. There are other changes in the global composition of political configurations. Liberal autocracies (which for much of the period forms the third largest category of regime type) have increased over time (somewhat waxing and waning throughout the 1972-1996 time period). By the 1992, Illiberal Democracies are about even with Liberal Autocracies as a percentage of all regime types, but despite the amount of attention they have received, they still only make up a relatively small number (e.g., 6 in 1992 within the sample observed as opposed to 15 Liberal Autocratic regimes in the same year). 18 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between Regardless of the number, it is clear by the end of this time period, however, that the distribution of polity characteristics is more complex than at any time since 1972 and that if one extends this to 1998 (which is not shown here) the distribution becomes even more equitable. This trend differs from the conclusions of Ward and Gleditsch (1997) who argue that the structure of the democracy (i.e., the rights component) has simplified overtime to a relatively equitable bifurcation between full autocracies and democracies. At the same time, this difference makes sense for they identify that the Freedom House indicator was more responsive to the participation aspect of political democracy (Ward and Gleditsch 1997, 380). This legitimates the measure. EXPLAINING CAUSAL RELATIONSHIPS: A SECOND LOOK While useful, the discussion above only provides an assessment of distributions and trends. One is still left wondering what factors account for position on the polity scale. Running multinomial logit regression models, prior expectations were largely supported, there are some interesting differences also identified. POLITY STRUCTURE WITH AUTOCRACY AS THE BASELINE From the results, the model (which is provided within Appendix 1a) explains 26 percent of the variance in the polity measure. Concerning the broader structural characteristics, the probability of being placed in the Liberal Democratic category is enhanced by economic development, as indexed by energy consumption per capita, as well as by population size.22 Modernization theory is thus supported within this analysis although interaction effects between these two variables might suggest different relationships as the excess burden of a large population without the economic wherewithal to contend with it is likely destabilizing, leading to other forms of governance. In this case, it likely perpetuates the probability of being placed in the Illiberal Autocratic category as no other regime type is influenced by such characteristics. 19 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between Addressing factors somewhat less impervious to human agency, below I explore other variables of interest. As presented within the figures, the y-axis represents the probability of a particular category being experienced and the x-axis represents the different values for the independent variables. The values themselves were selected from consideration of variable distributions (thus for example dichotomous variables are only examined for values when the condition of interest is present) as well as theoretical relevance. (Insert Figure 4a. About Here) Results are generally consistent with the brief sketches provided above, while at the same time displaying much greater complexity than suggested. The probability of placement within the Liberal Democratic category is enhanced by the overall amount of protest (near 85% at extremely high rates of occurrence), the length of time that the polity is around without change (near 20% at 30 years), and to a marginal extent political revolution (although increased frequency of this behavior exhibits a negative impact). It is plausible, therefore, to support the claim that (relative to Illiberal Autocracies - the base category) full democracies are differentiated by the continuous engagement of civil society within a stable political system. External factors also influence the likelihood of being placed within the Liberal Democratic category. Countries with fewer countries on their border are more likely to fall within this regime type, while the number of countries in the region increases the likelihood of placement. States that do not immediately have concerns for their safety or the protection of their borders and who have a multitude of potential contacts in their home region are more likely to be Liberal and Democratic. From the Illiberal Autocratic baseline, the two hybrids are not very comprehensible (i.e., not many variables are substantively important). While Polity Duration and Violent Conflict are statistically significant in their impact on Illiberal Democracy, they are ineffectual when one looks at the probabilities of being placed in this category, across different values of independent 20 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between variables. Placement almost appears random. Liberal Autocracies are slightly better accounted for. Interestingly, when one considers the constants, it is found that all other things being held constant (i.e., all independent variables being held at their mean), Liberal Autocracies are almost as likely as Liberal Democracies. Moreover, coups increase the probability of being in this category, while the Duration of the Polity tends to diminish this probability (albeit slightly). POLITY STRUCTURE WITH DEMOCRACY AS THE BASELINE In almost complete juxtaposition to the findings above, Illiberal Autocracies are found to thrive off of domestic and international strife – although not all forms.23 From the simulations (provided in Figure 4b below), one is able to see that at all levels of revolutionary activity, low levels of mass domestic conflict, and at most years of polity formation, these political systems are favored. This probability is subsequently increased when revolutionary activity worsens (i.e., continues) and decreased significantly when mass protest behavior expands in frequency as well as decreased slightly when the polity endures without change. Illiberal Autocracies decay through pressure exerted from below (as evidenced within the popular movements of 1989) as well as to a limited extent from internal dynamics (with its numerous organizational inefficiencies and diminished political legitimacy [Wintrobe 1998]). (Insert Figure 4b. About Here) Also different from the results above, it is clear that Illiberal Autocracies thrive within complex as well as contentious international contexts. The probability of these political systems is increased as the number of countries on the border rise as well as when the magnitude of conflict is worsened. The likelihood of Illiberal Autocracies is decreased, however, when the number of countries is increased, suggesting that certain aspects of the liberal thesis may be correct as the complexity of the external political environment seemingly compels political systems to adapt and open themselves internally in an effort to navigate through the exchanges that take place across societies. 21 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between As we move to our hybrid political systems, our capacity to explain polity structure is once again somewhat diminished. Illiberal Democracies are not well predicted by any variable. Across all statistically significant variables, when values are increased there is no increase in causal impact. This may bode well for those arguments that suggest that such regimes are fundamentally about negotiation, compromise and dynamic contingency – thus being less susceptible to this type of analysis. At the same time, this is uncertain because there has simply not been enough time for causal factors to either have their influence on polity configuration or for us to examine them rigorously in any great detail (hence the urgency contained within Zakaria’s [1997] article). Liberal Autocracies appear to be strongly associated with coups (three successful or attempted one’s result in a 33% probability of this category). Although not influenced to a dramatic extent, these systems also appear to thrive in relative contentious environments. As the magnitude of conflict within the region increases, the probability of Liberal Autocracy is increased. Mentioned above, this relationship may capture the attempt made by government authorities to maintain power within a potential crisis situation, while offering liberalization as an inducement to continue sacrificing democratic control and/or economic redistribution to non-military-related subjects. 4. A CONCLUSION AND AN AGENDA The present study attempted to shed some light on the neglected subject of polity structure. The impetus for this effort emerged from a relatively animated discussion about new and/or questionable forms of political systems. Here, I have attempted to identify a way that we might be able to better understand these diverse polities as well as initiate an examination of why the different forms exist. For this, I relied upon data drawn from numerous datasets over the 1972-1996 time period and across 164 Countries. From the results, certain inroads have been made with regard to our understanding of political systems. For example, the hybridic combinations of liberalism and democracy (i.e., 22 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between Illiberal Democracies and Liberal Autocracies) tend to be relatively small in number, although increasing, and not well explained by variables that work well in more traditional types of polities (i.e., Liberal Democracies and Illiberal Autocracies). Both hybrids tend to decrease as the institutions persist, suggesting that movement to one of the more traditional poles is undertaken over time. Additionally, some reaffirming insight into Liberal Democracy and Illiberal Autocracy has been provided. As anticipated, the probability of falling into the Liberal Democratic category is increased by mass protest behavior, political stability and the number of countries within the region, while the number of countries bordering a nation-state and revolutionary activity tends to decrease it. Illiberal Autocracies seem to falter within these same contexts. The latter sees to thrive when the society is new, convulsing with both internal as well as external conflict, and by an increased countries on the border. When the number of countries in the region increases, however, Illiberal Autocracies are decreased in likelihood. In certain respects, the present analysis is disappointing in that the insight into the hybrids came up somewhat short. There were no ”smoking guns” identified. In another respect, the analysis provides a rather sober conclusion that these hybrid systems are newly emerging as well as increasing in number (and thus worthy of attention), but still numerically in the minority, less susceptible to systematic analysis, and over time most likely to fall into the vortex of the Illiberal Autocratic-Liberal Democratic poles identified by Ward and Gleditsch (1997) – a pattern that has been operative for about 20 or 30 years. Additionally, the analysis provides an interesting beginning for further reflection about polity structure as it provides a measure that is predicted in a manner that we would expect (perhaps the best form of validity). Using this measure, five areas of additional investigation seem particularly worthwhile. (Insert Figure 5. About Here) First, the indicator of regime type developed should be further explored with regard to refining its’ sensitivity to various polity combinations. For example, Peru between 1990 and 23 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between 1996 manifested a Polity score of “7” but a civil liberty score of “3”, falling in the Liberal Autocracy category. Where should polities in this middle zone between extremes (or alternatively right at the boundaries of categories) be placed? Second, with the indicator one could rigorously begin to investigate the various paths to and from different political structures (e.g., like that suggested by Stepan [1986] or Karl [1990]). Considering Figure 5 (which represents a few examples of different political systems that have undergone change), it is clear that there are many trajectories that could be followed. The question is precisely why, how and how long?24 Third, it would be useful to explore diffusion, lagged as well as simultaneous effects, identifying whether or not more dynamic relationships exist. For example, where an initial opening of the polity structure increases the amount of mass protest that (in turn) leads to additional democratization or alternatively a rollback to autocracy. Additionally, it may be the case that multiple regimes undergo similar types of changes at the same time, thus increasingly the willingness of other states to experiment. Fourth, it is clear that the brief sketches of theoretical relationships provided earlier need to be developed so that causal processes can be more clearly identified. We must simply address Tilly’s (1998) comment with greater effort than that provided here. Finally, public opinion literature on tolerance and various indicators of trust in government and efficacy should be used in order to predict polity configuration as well as movement from one system to another. This would assist us in understanding what types of transitions individuals are ready for as well as when. Such an effort would overcome an important weakness in the literature where behavior and attitudes are kept distinct, but which normally rely upon one another to establish causal connections as well as theoretical/practical importance. As we continue to analyze the core and also the edges of democracy and more broadly polity configuration and understand the continual paradoxes of governance that emerge, efforts such as those identified above may invariably provide the best guide to improving our 24 Understanding Illiberal Democracies, Liberal Autocracies and Everything in Between understanding of where we have been and also providing the tools to identify where we may be going. 25