Latent territorial threat and democratic regime reversals

Abstract

Why do some democracies revert to non-democratic forms of governance? We develop an explanation of democratic reversals that emphasizes the influence of states’ external border relations on domestic politics. Latent threats to a state’s territory encourage political centralization of authority in the executive to defend against danger to the homeland. Latent territorial threat also facilitates the construction and maintenance of large land armies to fight threatening neighbors. Combined, latent territorial threat increases leaders’ domestic power, weakens democratic institutions, encourages other conditions threatening democratic survival, and, ultimately, leads to democratic reversals. Synthesizing prior research on territorial conflict, we generate a quantitative, continuous measure of latent territorial threat against all democracies with contiguous neighbors from 1946 to 2016, using Bayesian estimation. Empirical tests accounting for measurement uncertainty and other common determinants of reversals as well as brief reviews of individual cases of reversal provide robust evidence that democracy failed at higher rates in countries facing high levels of threats to their territory from neighbors. Our study implies that a complete account of the development of democratic institutions should emphasize that domestic factors alone fall short of explaining why democracies fail.

Keywords

democratic reversals domestic politics territorial threat

Why do democratic countries sometimes revert to autocratic governance? Many scholars have pointed out that democratic forms of government have an abundance of desirable economic, social, and international consequences.¹ As Mitchell (2012) argued, democracies had become so common and commanded so many of the world’s resources by the early 2000s that democratic norms set standards of behavior for both democracies and non-democracies alike.

Despite the putative benefits of democratic governance, however, democracies do not always survive once established. Democracies revert to non-democratic forms of governance with some regularity, with waves of democratization preceding ‘reverse waves’ of democratic backsliding (Waldner & Lust, 2018: 94). Why? Since World War II, domestic movements, international organizations, foreign aid donors, and international norms have facilitated transitions to democracy across the globe, which makes reversals to non-democratic systems puzzling since democratic systems and institutions remain preferred by majority populations.

This article argues that explanations of reversion need to account for the influence of democracies’ international environment. A country’s latent threat environment – in particular the threat to its territorial integrity from its neighbors – is an important but overlooked factor in determining which form of governance a country adopts and keeps (see Gibler & Tir, 2010; Gibler, 2012). This argument builds upon the longstanding logic that only those countries existing in relatively safe and peaceful environments can afford to democratize, while those living in threatening situations are likely to remain autocratic (Thompson, 1996; Tilly, 1985). We expand this logic to explain why democracies fail. Our argument suggests that countries with democratic institutions in threatening environments will face greater challenges to maintain these institutions and, consequently, have a higher risk of reversal.

This logic supplants prior work on democratic reversals. For example, just as democratic transitions are more likely for countries in areas with more democratic neighbors, democracies are less likely to survive in neighborhoods with fewer democracies (Gleditsch, 2002). Existing explanations for this clustering of democracies mostly focus on the regional diffusion of democratic norms (Elkins & Simmons, 2005) as well as the role of international organizations (Pevehouse, 2002; Donno, 2010) in promoting norms or bolstering domestic democratic institutions and processes. Conversely, in regions without strong democratic norms or established democracy-promoting organizations, reversals would be more likely. While these explanations provide credible theoretical and empirical accounts of regional patterns in democratic reversals and their prevention, they omit territorial threat as an important source of democratic reversal that has a more fundamental impact on regional environments.

A more complete understanding of obstacles to lasting democratization is especially important because mixed regimes – including those that democratized and then reverted – are particularly fragile and prone to internal violent conflict (Hegre et al., 2001). We show that democratization efforts may be less fruitful in the long run in countries facing latent territorial threat. We also know that governments facing low popularity among the public tend to ramp up diversionary territorial threats toward neighbors (Tir, 2010). Our study therefore contributes to the wider literature by showing that governments engaging in territorial diversion heighten the risk of democratic breakdown in neighboring countries, leading to a ripple effect.

Unlike previous territorial conflict research, we emphasize that territorial threat is a latent and slow-moving concept and, therefore, we develop a latent measure of territorial threats to the state. Using the new measure, and consistent with our argument, we show that latent territorial threat is associated with democracies sliding back into autocracy.

Why do democracies revert?

The literature on democratization and democratic consolidation has developed several explanations for democratic reversals. While acknowledging that the processes driving democratic reversals are complex (cf. Waldner & Lust, 2018), we briefly review a few key factors that feature particularly prominently in this literature to show that the impact of territorial threat on consolidation and reversals should be considered in conjunction with, or even prior to, existing explanations.

Domestic factors

Foundational for modernization theory, Lipset (1960) argued democracies with poorly performing economies are more likely to revert to autocratic institutions. Przeworski et al. (2000) found that wealthier democracies are far less likely to transition back to autocracy. Acemoglu & Robinson (2006) and others suggested that among economic factors, inequality and the resulting demand for redistribution are key drivers of both democratization and new democratic institutions’ prospects for survival. Although debates about the empirical support for this conjecture persist (e.g. Freeman & Quinn, 2012), the economy in broad terms, most commonly operationalized as economic development, is typically identified as one of the most powerful predictors of democratic survival (Gassebner, Lamla & Vreeland, 2013: 191).

Past experience with democratic transitions seems crucial in explaining whether new democratic institutions persist. Boix & Stokes (2003) found robust evidence for a strong association between previous democratic reversals and the increased odds of democratic breakdown. This association remains in the presence of economic variables (Boix & Stokes, 2003; Gassebner, Lamla & Vreeland, 2013) and implies that past reversals induce path-dependent patterns of future reversals. That reversals to autocracy tend to occur repeatedly within one country suggests an underlying factor driving these repeated reversals. We maintain that territorial threat is such an underlying factor, which can be linked both directly to reversals and also to some of the other domestic sources of reversals.

International factors

Reversals toward autocracy also occur in regional patterns. Here, the influence of neighboring countries as well as the role of international institutions, for instance through strengthening civil society or requiring democratic standards for accession, have emerged as robust correlates of democratic survival. The democracy diffusion and clustering literature (see e.g. Gibler & Tir, 2014; Gleditsch & Ward, 2006; Wejnert, 2005) has emphasized a number of theoretical processes behind the finding that new democracies with a larger number of democratic neighbors are more likely to remain democracies. Elkins & Simmons (2005), for instance, highlight two mechanisms behind the regional clustering of policies, including democratic consolidation. New democracies may adapt to altered conditions, specifically democratic standards in a region. Political actors in new democracies may also be increasingly likely to engage in learning processes that sustain democracy if a majority of proximate countries is democratic.

International institutions exert a democracy-stabilizing influence on member countries that recently democratized (Pevehouse, 2002). Hawkins (2008) suggests that international institutions can strengthen democracy by offering access to domestic nongovernmental organizations and civil society. Considering the central importance of elections and preventing election fraud for the survival of new democratic systems, Donno (2010) further illustrates the role of international factors in explaining democratic consolidation by emphasizing the role of international institutions in protecting the legitimacy and integrity of the electoral process.

Prior research thus suggests a candidate case of a country at high risk of democratic reversal. Such a country would be comparatively poor, have a short record of democratic institutions but a history of reversals to autocracy, be situated in a region with few other democracies, and not be deeply embedded in international organizations. This sketch leads to our main argument in this study: several of these risk factors of democratic reversals can be traced to a country’s relations with its neighbors over territory. Lingering territorial conflict can be linked to a number of political processes and variables, including democratization, regional democratic patterns, admission to democracy-sustaining IGOs, political centralization, mobilization, political intolerance, and even mass killing (see e.g. Hong & Kim, 2019). Some of these variables are central to the aforementioned risk factors for regime reversals as well as the more direct dynamics of regime reversals. Higher levels of territorial threat therefore destabilize democratic institutions and eventually increase the odds of democratic reversals.

Sources of short-lasting democratization

A natural question is why countries at risk of reversal become democratic in the first place. After all, territorial issues tend to be longstanding, with territorial claims persisting for decades (Huth & Allee, 2002), causing frequent militarized fights (Hensel, 1994), and even evolving into full-fledged enduring rivalries. Threatening environments also preclude democratization, according to works such as Thompson (1996) or Tilly (1985).

Nevertheless, some democratizations may occur despite threatening environments. Consistent with popular arguments in the comparative study of political systems (Lipset, 1960; Rueschemeyer, Stephens & Stephens, 1992), the literature on external influences on democratization assumes the actual transition to democracy to be a domestically driven process. When external threat is reduced, domestic actors can challenge the centralized structures for political decisionmaking and the authority of the military by demanding democratic reforms. This process, according to these arguments, ultimately results in democratization.

Throughout the past decades, however, a number of factors have contributed to democratization in potentially threatening environments. Domestically, economic pressure and distributive conflict (Acemoglu & Robinson, 2006), urbanization (e.g. Bates, 1981), and intra-elite conflicts (Haggard & Kaufman, 2012: 500) are key themes in the democratization literature. At least the first two factors are structural and can generate pressures for democratization even in the presence of territorial threat. These factors may induce unstable democratizations, even in states with high levels of territorial threat. Some of the key factors associated with democratization, especially during the second wave after World War II, can therefore apply even in countries facing high levels of territorial threat from neighbors.

Similarly, as Boix (2011), Haggard & Kaufman (2012), and many others suggest, external interveners, aid donors, and international organizations have each exercised a strong push for democracy at least in the ‘third wave’ of democratization from the 1970s through the early 1990s. Elements of this push include a pro-democracy discourse, tangible benefits such as promises of economic aid (Dunning, 2004), opportunities to accede to or associate with IGOs and large markets such as the European Union (Vachudova, 2005) and alliances such as NATO (Gheciu, 2005), and trade expansion that were conditioned on candidate countries’ progress toward democratization (Hafner-Burton, 2005).

How prevalent these factors, particularly international influences, are in cases of democratization can be seen in a closer look at the history of the cases of democratic reversal we investigate in this study (Table II). From Czechoslovakia in 1945 to the former Soviet states of Belarus and Armenia in 1991, more than a quarter of the cases entered or re-entered the state system as democratic governments. In most of these cases, democratic institutions governed in high-threat areas in which liberal governments could hardly be sustained according to our logic. Hence it is not surprising that seven of these 14 cases ended in military-led coups d’état. Of the remaining cases, many were replaced by right-wing governments with strong ties to the military.

We are agnostic toward the specific causes of initial democratization, but acknowledge that the aforementioned dynamics behind many, if not most, causes of democratization are not connected to territorial threat. That is, both structural internal and external pressures for democratization can operate regardless of whether a country has resolved territorial tensions with neighbors. Yet, territorial threats to the state continue following the transitions in these countries. Subsequently, these threats stall or undermine democratic reforms and increase the odds of regime reversal.

Consistent with this logic, we observe that the vast majority of reversals happened in countries with a short history of democratic institutions. As many as three-quarters of reversals happened less than ten years after transition to democracy, and two-thirds of reversal occurred in countries with fewer than ten years of experience with democracy since the country’s founding.² We argue in the next section that territorial threat is the determining factor in explaining the brevity of democratic institutions in these situations.

Territorial threat and democratic reversals

Border relations have domestic ramifications, from shaping resource allocation to creating focal points for political campaigns. We argue that these domestic consequences of relations with neighbors over territory are important for explaining cases of democratic reversals. Territorial threats exercise strong centralizing influence in the state, on popular opinion and the relative bargaining advantage of the executive vis-à-vis other government branches and the opposition. These forces translate into regime dynamics that encourage authoritarian forms of governance (Gibler, 2012).

One of these ramifications of threats to homeland territories is the centralization of the domestic politics of the state, across multiple levels of society. Threats alter the bargaining positions among elite groups. Opposition parties are more likely to back the government during crises. Scholars have long argued and observed that there is a rally-round-the-flag effect in public opinion particularly when crises encompass territorial threats (e.g. Mueller, 1973; Tir, 2010; Singh & Tir, 2018). Gibler (2010) suggests that, during crisis times, leaders have incentives to use the new-found political support from rallies to eliminate checks on their power. Opposition movements who challenge the leader can be cast as traitorous to the regime, which enables centralization of authority under the executive. It is also possible that general threats could be so salient that the opposition and general public willingly lend their support to the leader in order to persevere in the conflict and defend the state. In these instances, immediate survival trumps political freedom and contestation necessary for stable democracy.

The nature of territorial threats also changes the political environment in which domestic forces bargain and increases the influence of the military within the state. Unlike other forms of threat, disagreements over territory require the state to raise and sustain land-based militaries to take and hold state claims. Citizens of a country under territorial threat are more willing to fight on behalf of their country (Kim, 2020). Militarization is therefore more likely in the presence of territorial threat. A strong military adds another latent dimension to leader power. So long as the military is ‘protecting’ the state, though not actively engaged in conflict, opposition parties now must counter both popular support for defending the state and a strong military that can repress dissent. Challenges to leader policies are difficult when bargaining in the shadow of a strong, potentially repressive force.

There are further paths for how threat can hollow out democracy and thus reduce barriers to centralization and militarization. Hutchison (2014) and Tir & Singh (2010) show that attempts to carve out pieces of a state’s territory lead to a rise in political and social intolerance (see also Hutchison & Gibler, 2007), while territorial threat also undermines women’s welfare (Tir & Bailey, 2010). And Singh & Tir (2021) suggest that threat leads individuals to pay lip-service to political participation, rather than actually engaging in democracy-supporting behaviors.

In sum, states under territorial threat from neighbors are more likely to be institutionally centralized, to be dominated by nationalistic and intolerant mass attitudes, and to experience an increase in power of strong, standing militaries. Centralized authority enables leaders to prey upon industries and extract resources for their own gain. Experiencing periods of territorial threat should eventually leave states with poor economic growth prospects, with regimes that provide fewer public goods, and societies that experience high income inequality. In Olson’s (1993) terms, these are the regimes with stationary bandits that inhibit sustained democratization.

For our argument, the (separate) centralizing and militarizing effects of territorial threat imply that democratic institutions under conditions of high territorial threat are unlikely to last. As Svolik (2015) emphasized, democratic institutions can fail in two scenarios: incumbent leaders assume power and suspend democratic rule, or a coup d’état by the opposition begins a new regime. We suggest that territorial threat encourages both of these pathways to democratic reversals. Incumbent leaders in high-threat environments have incentives and opportunities to centralize authority and suspend democratic institutions. And aspiring leaders who do not hold elected offices, such as military generals or opposition politicians, may see opportunities to rise to power by circumventing the democratic process, in part citing extraordinary circumstances due to territorial threats to the homeland. Regular bargaining between elites and opposition movements – those forces that lead to democratic consolidation – is likely to be suspended or undermined under high territorial threat. All else equal, we therefore argue that democracy is unlikely to consolidate and survive in states that face persistent threats to their territory. In turn, this argument leads to the core hypothesis of our study:

Higher levels of latent territorial threat are associated with an increased likelihood of regime reversals in democratic states.

Estimating latent territorial threat

Understanding territorial threat as a background condition for domestic political processes corresponds to a conceptualization of threat as a latent and continuous variable, rather than being a binary condition. We accordingly develop a measure expressing the probability of engaging in militarized conflict with any neighboring state. Such conflict is the most direct manifestation of territorial threat to a state’s homeland territory because neighboring countries are the most likely source of territorial threat. Disagreements over the control of homeland territory are, meanwhile, rare between non-contiguous states. In other words, the political ramifications of centralization and militarization are most severe when a conflict credibly threatens a state’s territory. Conflicts with remote states do not carry the implications that drive changes in mass attitudes, political centralization, and mass militarization – which, we argue, increase the risk of democratic reversals.

This latent conceptualization of territorial threat distinguishes our study from previous work on the impact of territorial threat on political institutions that treated threat as a binary indicator: states were either under threat when a conflict was ongoing, or they were free from threat when conflict was absent. For instance, Thies (2005) and Lektzian & Prins (2008) found associations between threats through ongoing conflicts or rivalries and state capacity or centralization. Our argument emphasizes that territorial threat is continuous and moves slowly because the end of a militarized dispute does not indicate the absence of territorial threat and its consequences on democratic consolidation and reversal. And militarized interstate disputes rarely emerge out of nowhere, but are the culmination of previous tensions (see also Nieman & Gibler, forthcoming). Unlike binary indicators, our latent measure of territorial threat captures this underlying dimension of territorial threat. Observed disputes and territorial threat are distinct; some countries continue to experience high territorial threat after MIDs end. Figure A2 in the Online appendix illustrates this in more detail, and we show empirically that the latent operationalization makes a difference in this study.

Data

We identify a number of factors that scholars have associated with fatal militarized disputes (MIDs) between neighboring states. These factors contribute to an empirical model of territorial threat that produces a latent measure of territorial threat, while accounting for measurement uncertainty. Our conceptualization of territorial threat and its consequences requires a country to have at least one contiguous neighbor to experience a territorial threat to its homeland. We create this list from version 3.2 of the Correlates of War Direct Contiguity data (Stinnett et al., 2002) and include all dyad-years (1946–2016) of countries that share a land or river border.

We estimate the fatal militarized dispute propensities separately for each dyad-year in the data. Because the country-year is the unit of analysis for the investigation of democratic reversals, we aggregate different levels of territorial threat in cases where one country faces different threat levels from more than one neighboring country. Here, we assume that the highest territorial threat level is the most salient one that influences the dynamics outlined in the theoretical argument. Information about MIDs comes from the Militarized Interstate Dispute data version 4.3 (Palmer et al., 2020);³ further information on our data is in the Online appendix.

Predictors of territorial threat are selected based on (1) previous work on interstate conflict and (2) evaluating these variables empirically using a random forests classifier. They include variables capturing past interstate relations over territory and conditions that affect the occurrence of militarized disputes. A list of variables and explanation for their inclusion is provided in the Online appendix.

Predictive model

Because territorial threat manifests itself through the occurrence of fatal MIDs between neighboring states, the unit of analysis in this model is the dyad-year. And given that the purpose of this predictive model is to generate the best possible estimate of latent territorial threat that a state faced in given year, causal identification is not a concern.

To generate a latent measure that incorporates uncertainty, we fit a Bayesian regression model with a logit link function to the data. We obtain a posterior distribution of the latent territorial threat measure and later incorporate the uncertainty around the territorial threat estimate in our analysis of democratic reversals. The Online appendix contains more details on the specification and estimation of the Bayesian territorial threat model, including coefficient estimates in Table A3. All Bayesian estimates were obtained using Cauchy distributions centered around 0 as prior distributions. Given the goal of this model, coefficient estimates are of secondary importance, but their values are in line with prior research on militarized interstate disputes. As one accuracy measure, the area under the ROC curve is 0.93, which compares favorably with other studies predicting armed conflict (see e.g. Ward, Greenhill & Bakke, 2010).

To obtain a territorial threat measure at the country level, we use the highest territorial threat level across all of a country’s dyads in a given year. This results in a continuous measure ranging from 0 to 1, which we multiply by 100 to obtain threat scores in percentage points. For each country-year, our estimation yields a distribution of draws from the posterior estimates of territorial threat levels. We use this distribution to express measurement uncertainty around the concept of territorial threat, illustrated in Figure 1 and employed in our main model. Doing so is consistent with recent efforts to account for measurement uncertainty in a more systematic manner (see e.g. Karreth, 2018; McManus & Nieman, 2019).

Empirical test: Territorial threat and democratic reversals

We evaluate the hypothesis that higher levels of latent territorial threat are associated with a higher probability of democratic reversals. We then re-examine 52 cases of reversal since 1946 to better understand the causal processes that led to democratization originally and then its reversal.

Unit of analysis and spatial-temporal domain

The population of interest comprises countries at risk of democratic reversals: all countries that are commonly considered democratic. In line with prior research and the suggested classification of the creators of the Polity IV scale (Marshall & Jaggers, 2009), we consider countries ranked 6 or higher on the Polity IV $- 10$ to $10$ scale as democratic. This distinction is relevant, among other reasons, because prior work shows systematically different outcomes around this threshold (e.g. fewer civil wars in democracies: Hegre et al., 2001; Ward & Gleditsch, 1998). Because reversal presumes that a polity is democratic, we only include those country-years in the sample that the Polity project explicitly coded as democratic.

For theoretical and empirical reasons, democracies after World War II are the relevant population for our analysis. The international push toward democratization began after World War II, creating more democracies at risk of reversal. Reliable data on other predictors of reversals are consistently available after 1945 as well.⁴ The data used in the analysis cover 1946–2016 and include 2,972 country-years and 97 countries.

Figure 1.

Latent territorial threat for all 52 democratic reversals in the analyses, 1946–2016

Outcome: Democratic reversal

We classify all country-years where a country’s Polity IV score drops below 6 as incidents of democratic reversals. Because any classification other than democracy indicates a removal of democratic structures, we also consider transitions from democracy as reversals. By these rules, after 1945, 52 reversals occurred in 38 countries with contiguous neighbors, that is, countries subject to any level of territorial threat. Figure 1 lists the cases and Tables II and A1 offer additional details on them.

Predictor: Territorial threat

The original latent territorial threat variable based on our estimation is a probability that ranges from 0 to 100% risk in theory and 0% risk to 67% risk in our sample. In Figure 1, we illustrate the latent territorial threat levels of all 52 cases of democratic reversals in countries with contiguous neighbors. Over two-thirds of cases (36 out of 52) rank above the median of latent territorial threat of all democratic country-years during the period of interest. As an example, Turkey in 1980 has a latent territorial threat score of 35, implying a 35% probability that it will face a fatal MID with a neighbor in that year. In other words, one out of three cases with Turkey’s parameters would face a fatal MID in that year.

Because of the highly right-skewed distribution of the latent territorial threat variable (most countries face very low levels of territorial threat), we use two transformations to facilitate estimation and interpretation. The first measure is a logarithmic transformation of the latent territorial threat variable, averaged over two years to avoid distortions from spikes in the estimate. The second measure cuts the original score into ten deciles, or bins, where a country-year receives a value from 1 to 10 based on its percentile position on the latent territorial threat measure. The first bin contains the bottom 10% of country-years based on estimated territorial threat levels, and the last bin contains the 10% of country-years with the highest territorial threat levels. Each bin or decile contains several hundred observations. Another advantage of this measure is that it reduces the influence of rare outliers with particularly high values of territorial threat on the results.

Confounding variables

Our empirical model of democratic reversals adjusts for variables that a meta-study (Gassebner, Lamla & Vreeland, 2013) has found to be most robustly associated with democratic reversals. To avoid an oversaturated model, we follow their lead and draw specifically on these domestic factors as control variables, in addition to other predictors that we include to capture other international-level dynamics identified in our argument. Besides regional and global democracy trends, these include memberships in highly structured IGOs (Tir & Karreth, 2018); we also account for rivalries with non-contiguous states to capture a separate variant of external threat. A detailed discussion of these variables can be found in the Online appendix. As an alternative to regression adjustment, the Online appendix also reports results from regression on a matched sample.

Estimation

Our first model fully accounts for uncertainty in estimating latent territorial threat. This uncertainty carries over into estimates of the relationship between territorial threat and reversals, implying a conservative approach that has, in different forms, been applied elsewhere (e.g. McManus & Nieman, 2019). In each iteration of the Bayesian sampling process, every country-year observation is assigned a value of territorial threat that is a draw from a (normal) distribution with the mean and variance of our predictive model of territorial threat. The estimated relationship between territorial threat and democratic reversals then corrects for the fact that territorial threat is measured with uncertainty. In a second model, where we split observations into ten deciles of territorial threat, we treat the decile as the observed value for estimation purposes. All Bayesian estimates were obtained using Cauchy distributions centered around 0 as prior distributions.⁵

Our analysis makes the assumption that after controlling for other determinants of reversals, there are no direct pathways from correlates of territorial threat to reversals. The specification of the analysis of reversals in Table I takes precautions to sustain this assumption by capturing possible neighborhood influences other than territorial threat, including democracy and reversals in countries’ direct neighborhood and wider region.

Findings

In our analysis, democratic countries facing higher territorial threat are substantially more likely to revert to non-democratic forms of governance. Taking uncertainty into account (Model 1) and reducing the influence of high-threat outliers (Model 2), we find the statistical probability of such a difference in reversal risk between democracies at low versus high threat to exceed 95% (Table I and Figure 2).

Table I.

Posterior estimates: Territorial threat (including measurement uncertainty) and democratic reversals, 1946–2016

	(1) TT logged (with uncertainty)			(2) TT deciles
	Median	Std dev.	Pr(Estimate) $^{†}$	Median	Std dev.	Pr(Estimate)
Territorial threat	0.138*	0.089	95.2%	0.097*	0.058	95.5%
HSIGO memberships	–0.308	0.258	88.5%	–0.306	0.257	88.7%
Perc. democratic within 500 km	–0.247	0.217	87.1%	–0.243	0.218	87%
Reversals in region	–0.243	0.254	83.4%	–0.25	0.254	84.2%
Perc. democratic (global)	–0.303	0.427	76.1%	–0.281	0.432	74.7%
Post-Cold War	0.514	0.776	75.2%	0.481	0.775	74.2%
GDPpc (t–1, logged)	–0.353*	0.197	96.5%	–0.353*	0.198	96.4%
Polity (t–1)	–0.398*	0.194	98.1%	–0.393*	0.194	97.9%
Previous reversals	0.315*	0.185	95.2%	0.318*	0.184	95.4%
Years as democracy (logged)	–0.544*	0.217	99.4%	–0.545*	0.216	99.4%
Ethnic fractionalization	0.015	0.154	53.8%	0.017	0.155	54.3%
Non-contiguous rivalries	–0.099	0.252	66.4%	–0.095	0.252	65.7%
Intercept	–4.996*	0.677	100%	–6.062*	0.658	100%
Country-years	2,972			2,972

$*$ indicates a probability of 95% or higher that the relationship is in the direction of the median estimate.

$^{†}$ Pr(Estimate) is the share of the posterior distribution’s draws that are in the same direction as its median. See Figure 2 for illustration.

Cell entries summarize posterior draws from Bayesian logistic regression estimates.

Figure 2.

Posterior distribution of logit coefficient estimates for the territorial threat variables in Table I

To illustrate the substantive dimension of this relationship, consider numeric differences in the risk of democratic reversals. These reversals are rare; the baseline risk (at the lowest level of territorial threat) is about 1% such that one in a hundred typical democracies with no territorial threat would revert back to autocracy. This risk doubles, however, in a country with a territorial threat level of about 7% (in the 75th percentile of the measure) and it increases by a factor of 2.5 in a country at the 90th or higher percentile of territorial threat. The posterior distribution of these changes indicates that these differences are highly likely to be positive, with 95% or more of the posterior draws above 0.

Control variables generally perform as prior research leads us to expect. More consolidated democracies (indicated by higher polity scores, longer existence of democratic institutions, and fewer prior reversals) are less likely to revert. Wealthier countries are also less likely to revert. Estimates for other covariates, including regional dynamics, are noisier. We interpret these findings altogether as suggesting that territorial threat covers what might otherwise be an unobservable confounder in analyses of democratic reversals performed in other studies. This would also account for the less precise estimates around the impact of regional dynamics.

Latent, continuous threat versus ongoing disputes

This study considers territorial threat as a latent, continuous, and slow-moving concept rather than a direct and instantaneous consequence of the absence or presence of an observed dispute. Militaries engaged in active conflicts are busy fighting and have less of an impact on domestic politics. However, when those militaries return home or if the threat never turns violent, then those same militaries present themselves as powerful domestic forces within the state.⁶ Our empirical analysis supports this distinction. When we follow the practice of other studies (e.g. Gibler, 2010) and operationalize territorial threat as a state’s involvement in an ongoing militarized interstate dispute over territory, we find no association between threat and democratic reversals. The same applies to using a state’s involvement in a territorial MID in the previous year or in fatal MIDs in the present or previous year.⁷ This null finding is consistent with our argument: the impact of territorial threat on the social, political, and economic dynamics that bring about democratic reversals is a continual process that does not end or start with the onset of an observed militarized dispute.

Model fit and robustness checks

The models of democratic reversals classify observed reversals well. The Online appendix contains separation and ROC plots (Figures A7 and A8) that suggest that our models perform well in this respect. The Online appendix also reports more details on the following robustness tests, all of which returned similar or near identical results as reported above. (1) We match cases of reversals and non-reversals on potential confounders before estimating the relationship between latent territorial threat and reversals. (2) We use informed prior distributions on covariates, based on the literature. (3) We estimate a reduced model (with all available covariates) for the 1816–2016 time period.

Reviewing cases of democratization and reversal

Our argument also suggested that while territorial threat poses challenges for democratization, there are good reasons to expect instances of democratization even in states facing substantial territorial threats. In these cases, however, democratic institutions are typically short-lived. To probe this argument, we examine the 52 cases in our sample in which democratization was followed by reversal toward autocracy.⁸ This additional evidence serves the purpose of validating the logic of our argument. We find several patterns that explain why these states initially democratized, consistent with our argument above and despite high territorial threat. We group the cases into five types to facilitate this discussion and present them in Table II.

A plurality of our cases are states that came into existence as democracies. From Czechoslovakia in 1945 to the former Soviet states of Belarus and Armenia in 1991, more than a quarter of the cases (14 of the 52 we analyze) (re)entered the state system as democracies. In most of these first group cases, democratic governments were placed into high-threat areas; ten out of 14 of these cases had territorial threat levels exceeding the median value in the year prior to the reversal. In this environment, our argument suggests that liberal governments would be difficult to sustain. Hence, nine of the 14 cases ended in military-led coups d’état. Of the remaining cases, many were replaced by right-wing governments with strong ties to the military. Most of these cases tend to support the argument that external territorial threat leads to centralization and non-democratic government.

In a second set of cases, democracy was externally supported or even imposed onto the state; seven of the nine of these cases existed in environments where territorial threat was above the median value. For example, the United States pressured the leader of South Korea, Syngman Rhee, to step down in 1960, and supported elections. However, that democracy lasted only one year. Similar US-backed interventions occurred in Haiti twice (1990, 1994), and in the Dominican Republic in 1978.

Table II.

Background information on democratic reversals, 1946–2016

Group	Country	Reversal year	Latent territorial threat at reversal above median?	Years as democracy	Polity IV movement	Coup d’état by military?	Notes
Group 1: States that entered the system as democracies	Czechoslovakia	1947	Yes	2	–1 then –7	Following year
	Greece	1949	Yes	5	8 to 4
	Myanmar (Burma)	1962	Yes	14	8 to –6	Yes
	Nigeria	1966	No	6	7 to –7	Yes
	Uganda	1966	Yes	4	7 to 0	Self-coup
	Sierra Leone	1967	No	6	6 to –7	Yes
	Somalia	1969	Yes	9	7 to –7	Yes
	Malaysia	1969	Yes	12	10 to 1 to 4
	Lesotho	1970	No	4	9 to –9
	Bangladesh	1974	Yes	2	8 to –2	Yes
	Ukraine	1993	Yes	2	6 to 5
	Gambia	1994	No	29	8 to –7	Yes
	Belarus	1995	Yes	4	7 to 0
	Armenia	1995	Yes	4	7 to 3
Group 2: Externally driven democracy or reversal	Syria	1958	Yes	4	7 to –99		Joined Nasser’s Egypt in union
	Laos	1960	Yes	2	8 to –1	Yes	US-led efforts led elected, right-wing gov’t overthrown by coup
	South Korea	1961	Yes	1	–4 to 8 to –7	Yes	US asked Rhee to step down, supported elections, replaced by coup
	Dominican Republic	1963	No	1	–3 to 8 to 0	Yes	US/CIA against elected government, riots/strikes led to elected junta
	Uruguay	1971	Yes	19	8 to 3		US-led anti-left forces rigged election with Brazil; suspended gov’t
	Chile	1973	Yes	9	6 to –7	Yes	US-aided coup
	Haiti	1991	Yes	1	7 to –7	Yes	Military ruler left country at urging of US
	Dominican Republic	1994	No	16	6 to 5		Strong support for democracy from US
	Haiti	1999	Yes	5	7 to 2		US forced out military ruler with threat of int’l force
Group 3: ‘Caretaker militaries’ where democracy is allowed and then revoked	Pakistan	1958	Yes	2	8 to –8	Yes	Military-led elections in 1956, followed by coup in 1958
	Turkey	1980	Yes	7	9 to –5	Yes	Military allowed elections then overthrown by military coup
	Honduras	1985	Yes	3	6 to 5		Military negotiation of a coalition change
	Peru	1992	Yes	12	7 to 8 to –3 to 1	Yes	Military leadership agreed to a Constituent Assembly
	Niger	1996	Yes	4	8 to –6	Yes	Military-led transition to democracy
	Lesotho	1998	Yes	5	8 to 0		Military-led transition to democracy during interregnum period
	Malawi	2001	No	7	6 to 4		Military sided with demonstrators; food crisis and aftermath
	Thailand	2006	Yes	14	9 to –5 to –1	Yes
	Bangladesh	2007	Yes	16	6 to –6 back to 5	Partial
	Niger	2009	No	5	6 to –3 to 3 to 6	Yes
	Guinea-Bissau	2012	No	7	6 to 1	Yes	Coup in 2003 led to elections; 2012 coup replaced that gov’t
	Thailand	2014	Yes	3	7 to –3	Yes	Military coup replaces gov’t with opposition candidate
Group 4: Strong military states with factional fighting	Argentina	1976	Yes	3	6 to –9	Yes	Coup after violent repression, disappearances
	Pakistan	1977	Yes	4	8 to –7	Yes	Military coup
	Ghana	1981	No	2	6 to –7	Yes	Military coup that was signaled at start of democracy
	Nigeria	1984	Yes	5	7 to –7	Yes	Military coup; military strong during democracy
	Pakistan	1999	Yes	11	7 to –6	Yes	Military decapitated in plane crash, elections, then coup
	Mali	2012	No	20	7 to 0 to 5	Yes	Military sponsored democratic transition initially
	Burundi	2015	No	10	6 to –1	Yes	Failed coup, repressive response
Group 5: Democratic governance failures	France	1958	Yes	12	10 to 5	Partial	Unrest over Algeria campaign led to new gov’t and constitution
	Turkey	1971	Yes	11	8 to –2	Yes	Civil unrest led to military intervention
	Zambia	1996	Yes	5	6 to 1	Yes	Military unable to quell protests
	Nepal	2002	No	3	6 to –6		Maoist insurgency
	Venezuela	2006	Yes	48	6 to 5		Small initial move, with trend toward autocracy
	Russia	2007	Yes	7	6 to 4		Small initial move, with trend toward autocracy
	Malaysia	2014	No	6	6 to 5		Small initial move
	Ukraine	2014	No	20	6 to 4		Civil unrest, repressive response, then secessionist conflict
	Turkey	2014	Yes	31	9 to 3 to –4		President consolidates power after election; severe autocratic measures two years later
	Niger	2016	Yes	5	6 to 5	No	President re-elected while opposition boycotts election

Support for Haitian democracy led to one year of a liberal regime until a coup in 1991, and factional fighting continued throughout the democracy that emerged in 1994, culminating in President Préval’s dismissal of the Chamber of Deputies and much of the Senate five years later. Dominican democracy was more successful, lasting 16 years until an incumbent engaged in widespread electoral fraud to block a populist challenger. These are the cases of US-led support for democracy, but other external interventions were anti-democratic. In Laos (1960), in the Dominican Republic (1963), in Uruguay (1971), and in Chile (1973), the United States played a major role in fostering the reversal of democratic governments. Overall, these cases imply that a full model of democratization and democratic consolidation should incorporate external influences on both democratization and reversals to authoritarian institution.

We can group the third and fourth sets of reversals together as cases of factional fighting in states with strong militaries. Nevertheless, we differentiate between two types of reversals because of the relative power of the military compared to other factions in each case. In Group 3 of Table II, 13 democracies emerged in which the military allowed democratization in response to popular calls for democratic government. Ten of the eventual 12 reversals took place in an environment exceeding the median territorial threat value. In nine of the 12 cases, the military then overruled the democratic governments and replaced their rule through a military-led coup d’état. And in the other three cases, the militaries were strong in the wake of threats to the state, but the military also had a stake in society. The military negotiated among factions in Honduras in 1985, helped Lesotho transition during an interregnum period in 1998, and helped Malawian protesters against government actions in 2001. Our argument predicts centralized government and a strong military, and these types of military-supported transitions are consistent with that argument.

Also consistent with our argument are the cases in Group 4 in which strong militaries engaged in factional fighting within the state; five of the eight cases existed in an environment where the territorial threat was above the median. Each of these cases were relatively short-lived democracies in which one or more factions were able to bring about a representative government. Nevertheless, each of these cases ended in coups led by the military. The only difference between these cases and the authoritarian states we predict with the general model is that these countries had other forces that the military had to bargain with or fight against. The military faction was not strong enough to dominate the state, even though it had enough power to stage coups.

The final set of cases can best be characterized as democratic governments that either do not perform well or do not withstand moves by authoritarian-bent elites within the state. The military in Turkey (1972), for example, imposed its own government following widespread unrest and economic crisis in the state. Similar crises occurred in the Sudan (1989) and Zambia (1996), and each was followed by a military coup. Last, we find three cases of creeping authoritarianism in Nepal (2002), Venezuela (2006), and Russia (2007). Each of these countries may have had well-positioned or strong militaries within the state, but the reversals occurred more slowly than in the other cases. Territorial threat and its consequences still weakened democratic governments and empowered authoritarian forces. Seven out of ten of these cases had territorial threat value above the median.

This section presented a brief account of why cases of reversals democratized in the first place and how they reverted back to authoritarian institutions. In 71% of the cases, territorial threat environments exceeding the median value. Overall, we find anecdotal support consistent with our argument. The reversals since 1946 almost all had strong militaries when territorial threats were high. Many reverted from democracy through coups d’état. Those that did not experience coups had militaries that effectively controlled the choice over the type of government or had liberal governments facing uphill battles against authoritarian forces, the latter empowered by territorial threat. This evidence suggests that a high territorial threat environment, through the empowerment of the military within the state and the weakening of liberal governments, heightens the risk of reversals to autocratic institutions.

Conclusion

Debates about the primacy of country-level economic, political, or cultural determinants of democratic consolidation have long been prominent in both academic and policy circles. Others have emphasized the importance of external and regional consolidation dynamics, pointing to the diffusion of democracy, regional clustering, and the influence of international organizations. Our argument and findings emphasize that external factors are indeed central to explaining why democracies fail, but we highlight that external threats to a country’s territory are crucial for the consolidation and survival of democratic political institutions. We construct a latent, continuous measure of external territorial threat to test whether heightened threat indeed hinders democratic consolidation and increases the risk of reversals to autocratic institutions. Using this measure, we find a substantial impact of territorial threat on democratic reversal. The latent danger of facing militarized conflict with neighbors gives rise to political dynamics that undermine democratic institutions. Our study presents robust evidence that democracy is more likely to fail in countries that face high levels of threats to their territory from neighbors. While democratic reversals are rare, countries that do revert to non-democratic forms of governance face high levels of territorial threat compared to those that endure.

A concrete example of this dynamic today includes Ukraine. Facing a high probability of continued territorial conflict with Russia, Ukraine’s democratic institutions have seen multiple challenges since independence and especially in recent years. Until continued high levels of territorial threat emanating from the conflict with Russia subside, this study suggests that democratic consolidation in Ukraine is unlikely. As a counter-example, our results imply that countries taking steps to reduce territorial threat (per settling territorial disputes) stand higher chances of democratic consolidation. North Macedonia’s recent agreement with Greece thus brings good news for its democratic institutions.

From a broader policy perspective, our findings suggest caution about democratization trends in environments where territorial threat levels persist. Of the reversals we identify in this study, most occur not only under high levels of territorial threat, but also in cases where democratic institutions emerged due to more or less direct involvement of external actors. In such scenarios, democratic institutions are considerably less likely to persist compared to regions where territorial threat is low or absent. For the efforts of third-party states or international organizations, this trend implies that successfully mediating and resolving territorial tensions should take precedence over democracy promotion with a strong internal focus. On the flipside, our study suggests that when facing limited resources, states and organizations pursuing democracy promotion should focus on those countries that are under low territorial threat or that recently took steps to reduce existing territorial threats. North Macedonia is a good candidate case for such work.

For scholarship on democratization and democratic survival, our study implies that a complete account of the development of democratic institutions should emphasize that domestic factors alone fall short of explaining why democracies fail. Rather, some of the more important domestic predictors of democratic survival are themselves conditional on countries’ external threat environment. Future research might investigate whether similar dynamics apply to some of the external factors commonly associated with democratic survival as well, foremost the role of international organizations. Based on the findings of this study, it is likely that precisely the formation, membership, and effectiveness of the organizations typically associated with democratic stabilization are themselves contingent on low levels of territorial threat. Domestic factors commonly associated with democratic reversals may also be contingent on territorial threat, rather than exogenous influences on reversal. For instance, economic growth, redistribution, and the institutionalized division of political power are all more likely under low levels or in the absence of territorial threat. Future research should explore in more detail how these factors evolve in relationship to territorial threat.

Linking our findings with works suggesting a decline in warfare and territorial disputes, there may be reason for limited optimism about democratic consolidation. Declines in territorial threat, illustrated also in Figure A2 (Online appendix), do not only save lives, but they also suggest per our findings that there should be fewer externally driven democratic breakdowns in the future. This optimistic conclusion is at the same time limited by recent trends showing an uptick in territorial disputes around expansionist foreign policies.

Footnotes

Replication data

Replication data and Online appendix for this article can be found at https://www.prio.org/JPR/Datasets/ and . All analyses were conducted using R version 4.0.3.

Acknowledgements

We thank the editor and anonymous reviewers for helpful feedback. The authors contributed as follows: JK, JT, and DG conceptualized the study; JK compiled and analyzed the data; DG and JK compiled background information on cases; JK, JT, and DG wrote the article.

ORCID iD

Johannes Karreth

Jaroslav Tir

Douglas M Gibler

Notes

References

Acemoglu

Daron

Robinson

James A

(2006) Economic Origins of Dictatorship and Democracy. New York: Cambridge University Press.

Bates

Robert H

(1981) Markets and States in Tropical Africa: The Political Basis of Agricultural Policies. Berkeley, CA: University of California Press.

Boix

Carles

(2011) Democracy, development, and the international system. American Political Science Review 105(4): 809–828.

Boix

Carles

Stokes

Susan C

(2003) Endogenous democratization. World Politics 55(4): 517–549.

Donno

Daniela

(2010) Who is punished? Regional intergovernmental organizations and the enforcement of democratic norms. International Organization 64(4): 593–625.

Dunning

Thad

(2004) Conditioning the effects of aid: Cold War politics, donor credibility, and democracy in Africa. International Organization 58(2): 409–423.

Elkins

Zachary

Simmons

Beth A

(2005) On waves, clusters, and diffusion: A conceptual framework. Annals of the American Academy of Political and Social Science 598(1): 33–51.

Freeman

John R

Quinn

Dennis P

(2012) The economic origins of democracy reconsidered. American Political Science Review 106(1): 58–80.

Gassebner

Martin

Lamla

Michael J

Vreeland

James Raymond

(2013) Extreme bounds of democracy. Journal of Conflict Resolution 57(2): 171–197.

10.

Gheciu

Alexandra

(2005) Security institutions as agents of socialization? NATO and the ‘New Europe’. International Organization 59(4): 973–1012.

11.

Gibler

Douglas M

(2010) Outside-in: The effects of external threat on state centralization. Journal of Conflict Resolution 54(4): 519–542.

12.

Gibler

Douglas M

(2012) The Territorial Peace: Borders, State Development, and International Conflict. New York: Cambridge University Press.

13.

Gibler

Douglas M

Tir

Jaroslav

(2010) Settled borders and regime type: Democratic transitions as consequences of peaceful territorial transfers. American Journal of Political Science 54(4): 951–968.

14.

Gibler

Douglas M

Tir

Jaroslav

(2014) Territorial peace and democratic clustering. Journal of Politics 76(1): 27–40.

15.

Gibler

Douglas M

Miller

Steven V

Little

Erin K

(2017) An analysis of the militarized interstate dispute (MID) dataset, 1816–2001. International Studies Quarterly 60(4): 719–730.

16.

Gleditsch

Kristian Skrede

(2002) All International Politics is Local: The Diffusion of Conflict, Integration, and Democratization. Ann Arbor, MI: University of Michigan Press.

17.

Gleditsch

Kristian Skrede

Ward

Michael D

(2006) Diffusion and the international context of democratization. International Organization 60(4): 911–933.

18.

Hafner-Burton

Emilie M

(2005) Trading human rights: How preferential trade agreements influence government repression. International Organization 59(3): 593–629.

19.

Haggard

Stephan

Kaufman

Robert R

(2012) Inequality and regime change: Democratic transitions and the stability of democratic rule. American Political Science Review 106(3): 495–516.

20.

Hawkins

Darren

(2008) Protecting democracy in Europe and the Americas. International Organization 62(3): 373–403.

21.

Hegre

Håvard

Ellingsen

Tanja

Gates

Scott

Gleditsch

Nils Petter

(2001) Toward a democratic civil peace? Democracy, political change, and civil war, 1816–1992. American Political Science Review 95(1): 33–48.

22.

Hensel

Paul R

(1994) One thing leads to another: Recurrent militarized disputes in Latin America, 1816–1986. Journal of Peace Research 31(3): 281–297.

23.

Hong

Mi Hwa

Kim

Nam Kyu

(2019) How do external territorial threats affect mass killing? Journal of Peace Research 56(4): 529–544.

24.

Hutchison

Marc L

(2014) Tolerating threat? The independent effects of civil conflict on domestic political tolerance. Journal of Conflict Resolution 58(5): 796–824.

25.

Hutchison

Marc L

Gibler

Douglas M

(2007) Political tolerance and territorial threat: A cross-national study. Journal of Politics 69(1): 128–142.

26.

Huth

Paul K

Allee

Todd L

(2002) The Democratic Peace and Territorial Conflict in the Twentieth Century. Ann Arbor, MI: University of Michigan Press.

27.

Karreth

Johannes

(2018) The economic leverage of international organizations in interstate disputes. International Interactions 44(3): 463–490.

28.

Kim

Nam Kyu

(2020) Territorial disputes and individual willingness to fight. Journal of Peace Research 57(3): 406–421.

29.

Lektzian

David

Prins

Brandon C

(2008) Taming the leviathan: Examining the impact of external threat on state capacity. Journal of Peace Research 45(5): 613–631.

30.

Lipset

Seymour Martin

(1960) Political Man: The Social Bases of Politics. Garden City, NY: Doubleday.

31.

Marshall

Monty G

Jaggers

Keith

(2009) Polity IV dataset. Center for International Development and Conflict Management, University of Maryland.

32.

McManus

Roseanne W

Nieman

Mark D

(2019) Identifying the level of major power support signaled for protégés: A latent measure approach. Journal of Peace Research 56(3): 364–378.

33.

Mitchell

Sara McLaughlin

(2012) Norms and the democratic peace. In: Vasquez

John A

(ed.) What Do We Know About War? 2nd edn. Lanham, MD: Rowman & Littlefield.

34.

Mueller

John D

(1973) War, Presidents and Public Opinion. New York: Wiley.

35.

Nieman

Mark D

Gibler

Douglas M

(forthcoming) Peaceful neighborhoods and democratic differences. Journal of Politics.

36.

Norris

Pippa

(2012) Making Democratic Governance Work: How Regimes Shape Prosperity, Welfare, and Peace. New York: Cambridge University Press.

37.

Olson

Mancur

(1993) Dictatorship, democracy, and development. American Political Science Review 87(3): 567–576.

38.

Palmer

Glenn

D’Orazio

Vito

Kenwick

Michael R

McManus

Roseanne W

(2020) Updating the militarized interstate dispute data: A response to Gibler, Miller, and Little. International Studies Quarterly 64(2): 469–475.

39.

Pevehouse

Jon C

(2002) With a little help from my friends? Regional organizations and the consolidation of democracy. American Journal of Political Science 46(3): 611–626.

40.

Przeworski

Adam

Alvarez

Michael E

Cheibub

Jose A

Limongi

Fernando

(2000) Democracy and Development. New York: Cambridge University Press.

41.

Rueschemeyer

Dietrich

Stephens

Evelyne Huber

Stephens

John D

(1992) Capitalist Development and Democracy. Chicago, IL: University of Chicago Press.

42.

Singh

Shane P

Tir

Jaroslav

(2018) Partisanship, militarized international conflict, and electoral support for the incumbent. Political Research Quarterly 71(1): 172–183.

43.

Singh

Shane P

Tir

Jaroslav

(2021) Threat-inducing violent events exacerbate social desirability bias in survey responses. American Journal of Political Science (https://doi.org/10.1111/ajps.12615).

44.

Stinnett

Douglas M

Tir

Jaroslav

Schafer

Philip

Diehl

Paul F

Gochman

Charles

(2002) The Correlates of War Project direct contiguity data, version 3.0. Conflict Management and Peace Science 19(2): 59–67.

45.

Svolik

Milan W

(2015) Which democracies will last? Coups, incumbent takeovers, and the dynamic of democratic consolidation. British Journal of Political Science 45(4): 715–738.

46.

Thies

Cameron G

(2005) War, rivalry, and state building in Latin America. American Journal of Political Science 49(3): 451–465.

47.

Thompson

William R

(1996) Democracy and peace: Putting the cart before the horse? International Organization 50(1): 141–174.

48.

Tilly

Charles H

(1985) War making and state making as organized crime. In: Evans

Peter B

Rueschemeyer

Dietrich

Skocpol

Theda

(eds) Bringing the State Back In. New York: Cambridge University Press, 169–191.

49.

Tir

Jaroslav

(2010) Territorial diversion: Diversionary theory of war and territorial conflict. Journal of Politics 72(2): 413–425.

50.

Tir

Jaroslav

Bailey

Maureen

(2010) Painting too ‘Rosie’ a picture: The impact of external threat on women’s economic welfare. Conflict Management and Peace Science 35(3): 248–262.

51.

Tir

Jaroslav

Karreth

Johannes

(2018) Incentivizing Peace: How International Organizations Can Help Prevent Civil Wars in Member Countries. New York: Oxford University Press.

52.

Tir

Jaroslav

Singh

Shane P

(2010) Get off my lawn: Territorial civil wars and subsequent social intolerance in the public. Journal of Peace Research 52(4): 478–491.

53.

Vachudova

Milada Anna

(2005) Europe Undivided: Democracy, Leverage, and Integration since 1989. New York: Oxford University Press.

54.

Waldner

David

Lust

Ellen

(2018) Unwelcome change: Coming to terms with democratic backsliding. Annual Review of Political Science 21(1): 93–113.

55.

Ward

Michael D

Gleditsch

Kristian Skrede

(1998) Democratizing for peace. American Political Science Review 92(1): 51–61.

56.

Ward

Michael D

Greenhill

Brian D

Bakke

Kristin M

(2010) The perils of policy by p-value: Predicting civil conflicts. Journal of Peace Research 47(4): 363–375.

57.

Wejnert

Barbara

(2005) Diffusion, development, and democracy, 1800–1999. American Sociological Review 70(1): 53–81.