Z Vgl Polit Wiss (2016) 10:241–272 DOI 10.1007/s12286-016-0312-y AUFSÄTZE

The contextualized index of statehood (CIS): assessing the interaction between contextual challenges and the organizational capacities of states Oliver Schlenkrich · Lukas Lemm · Christoph Mohamad-Klotzbach

Published online: 1 December 2016 © Springer Fachmedien Wiesbaden 2016

Abstract Although the measurement of the quality of statehood has become an important research field in Comparative Politics, most of the currently used indices are flawed by major weaknesses. To address such weaknesses, we developed the “Contextualized Index of Statehood” (CIS). This index is based on the two most essential dimensions of statehood: “monopoly on the use of physical force” and “administration”. Applying a new aggregation method called “variable threshold”, we highlight the interaction between the organizational capacities of a state and the contextual challenges it may face in order to measure the quality of statehood. First, we demonstrate the concept, measurement and aggregation. Then we investigate the validity of our index before presenting our empirical findings, including a modelbased cluster analysis. Keywords State fragility · Measurement of the quality of statehood · Model-based clustering · Failed states · Contextual challenge · Organizational capacity · Variable threshold

O. Schlenkrich () · L. Lemm, B.A. · C. Mohamad-Klotzbach, M.A. Institut für Politikwissenschaft und Soziologie, Lehrstuhl für Vergleichende Politikwissenschaft und Systemlehre, Universität Würzburg, Wittelsbacherplatz 1, 97074 Würzburg, Germany E-Mail: [email protected] L. Lemm, B.A. E-Mail: [email protected] C. Mohamad-Klotzbach, M.A. E-Mail: [email protected]

K

242

O. Schlenkrich et al.

Ein Kontextualisierter Index der Staatlichkeit (KIS): Zur Interaktion zwischen kontextuellen Herausforderungen und organisationalen Kapazitäten des Staates Zusammenfassung Obwohl die Messung von Staatlichkeit zum wichtigen Forschungsgebiet innerhalb der Vergleichenden Politikwissenschaft wurde, zeigen aktuell verwendete Indizes starke Defizite. Deshalb haben wir den „Kontextualisierten Index der Staatlichkeit“ (KIS) entwickelt. Der Index basiert auf den zwei essentiellsten Dimensionen der Staatlichkeit: „Physisches Gewaltmonopol“ und „Administration“. Unter Verwendung des neuen Aggregationsverfahrens des „variablen thresholds“ analysieren wir die Interaktion von organisationalen Kapazitäten des Staates und kontextuellen Herausforderungen zur Messung der Qualität von Staatlichkeit. Zuerst stellen wir Konzept, Messung und Aggregation dar; danach untersuchen wir die Validität des Index, um schließlich empirische Befunde inklusive einer modellbasierten Cluster-Analyse zu präsentieren. Schlüsselwörter Staatszerfall · Messung von Staatlichkeit · Modell-basierte Cluster-Analyse · Kontextuale Herausforderungen · Organisationale Kapazitäten · Variabler threshold

1 Introduction Neither the term nor the debate on state fragility is new. A few years after the demand to “bring the state back in” (Evans et al. 1985, p. 317), the end of the cold war and the resultant cancelling of foreign assistance revealed the dysfunctioning of numerous states, emphasizing the relevance of statehood as an analytical concept for comparative research. Although in the 1990 s a number of scholars discussed the problem of the so-called “failed” or “collapsed” states (Helman and Ratner 1992; Gros 1996; Zartman 1995), academic research into state fragility really took off as a result of 9/11 (Beisheim and Schuppert 2007; Brock et al. 2012; Büttner 2004; Lambach 2008; Rotberg 2003, 2004a; Schneckener 2006; Straßner and Klein 2007; Weiss and Schmierer 2007). Initially, the influence of parallel policy discussions showed a potential to narrow the research perspectives (Lambach and Bethke 2012, pp. 11–12), but nowadays academic institutionalization lies ahead of the field of research, as is indicated by the numerous publications on various aspects of statehood as well as the establishment of collaborative research centers.1 One important issue in this debate is the way the quality of statehood is measured. Although in the recent years various measurements of state fragility have been developed (e. g. Call 2011; Grävingholt et al. 2012; The Fund for Peace 2015; Rice and Patrick 2008, Marshall and Cole 2014), most of these indices are flawed by

1

See for example Berlin (http://www.sfb-governance.de/) and Bremen (http://www.sfb597.uni-bremen. de/).

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

243

major weaknesses2: First, their concept of “statehood” is often too broad, meaning they include dimensions or components which should be better understood as causes of state fragility, rather than as a part of the narrow definition of statehood; or they incorporate regime characteristics into their definition of statehood – failing to differentiate between the quality of statehood and the quality of democracy. Second, they focus on the outcome of state fragility rather than on the state itself, for example its organizational capacities. Third, they fail to highlight the contextual constraints (e. g. natural disasters, rebel groups, organized crime) with which a state has to deal, or they mix them with other indicators which show little added value (e. g. The Fund for Peace 2015). So, how can the quality of statehood be more accurately assessed? In this article, we present the “Contextualized Index of Statehood” (CIS) which covers 117 developing and transformation countries from 2006 to 2012 on a two-year basis. The index is based on what we regard as the two most essential dimensions of statehood: the monopoly on the use of physical force and administration. For these two dimensions, we explicitly include not only a state’s organizational capacities but also the challenges it faces as contextual factors. In order to represent these interactions we additionally introduce a new method of aggregation which we call the “variable threshold”. We begin by discussing the conceptualization, measurement and aggregation of the CIS (2).3 Then we investigate its validity (3) before presenting our empirical findings (4). A summary is provided in the last chapter (5).

2 Framework of the “contextualized index of statehood” 2.1 Concept: the dimensions “monopoly on the use of physical force” and “administration” versus their contextual challenges

We define empirical statehood as the functioning of two essential dimensions: “monopoly on the use of physical force” and “administration”. We regard the monopoly on the use of physical force as functioning if it can successfully deploy its organizational capacities (e. g. military and police force) to overcome competitors (e. g. rebel groups, organized crime) who threaten the state’s monopoly on the use of physical force. The administration is considered functioning if the state establishes an administrative structure which allows it to deliver basic goods despite contextual challenges (e. g. landlocked countries, natural disasters). Both dimensions have equal weight, although the functioning of the monopoly on the use of physical force is a precondition of the functioning of the administration. But a state

2

For a good overview of strength and weaknesses of some of these indices see Bethke (2012), Mata and Ziaja (2009), and Ziaja (2012). 3 The steps of our presentation are based on the scheme for evaluating indices by Munck and Verkuilen (2002) (for a detailed discussion see Müller and Pickel 2007; Pickel et al. 2015). Even though they only refer to indices of democracy, these schemes are also usable for the description of indices of fragile states (Ziaja 2012).

K

244

O. Schlenkrich et al.

has only steering capability when its administration is functioning as well. What does this definition imply? First of all, it is useful to distinguish between juridical and empirical statehood (Jackson and Rosberg 1982; Jackson and Sørensen 2007, p. 18–25). Whereas juridical statehood refers to the legal status of a state, its being viewed as sovereign by other states, empirical statehood refers to the actual quality of the institutions and organizational capacities of a state. Thus, it is possible that an entity has empirical statehood, while it lacks juridical statehood (“de facto states”). It is also possible that a state has juridical statehood, even though it lacks empirical statehood (“quasistates” resp. “failing states”).4 Second, we use only a two-dimensional approach, although newer concepts of empirical statehood refer to the functioning of three dimensions, namely security, capacity, and legitimacy (Schneckener 2004; Call 2011; Grävingholt et al. 2012; for a discussion see Bethke 2012). However, as we will point out, these concepts are often too maximalist. In contrast, we wish to focus on a more narrowly defined concept of “statehood” – especially in distinction to the concept of “regime”5 – in order to avoid conceptual stretching (Sartori 1970) and to separate the definitional aspects from necessary, obstructive or promotional factors (Lauth 2015, pp. 6–7). The security function is usually understood as a “state’s prime function [...] to prevent cross-border invasions and infiltrations, and any loss of territory; to eliminate domestic threats or attacks upon the national order and social structure; to prevent crime and any related dangers to domestic security” (Rotberg 2004b, p. 6). While we agree that a state has to protect its citizens from other non-state forces by controlling its territory in order to maintain the monopoly on the use of physical force, we disagree with concepts which focus on the idea of “human security” and which posit, normatively and morally, that a state must also protect its citizens from itself (Schneckener 2004, p. 13; Englehart 2009). If this concept is introduced into the definition of statehood, it blurs the analytical distinction between state and regime. Instead, regime characteristics will decide how a state uses its force against its own citizens (Merkel 2010, p. 64). In fact, a state which can intimidate its citizens can be seen as having strong capabilities. This means that both a democratic and an autocratic state can have a high quality of statehood.6 Accordingly, we do not call this dimension “security”; instead, we use – like the Bertelsmann Stiftung (2014a, p. 124) – the more neutral term “functioning of the monopoly on the use of physical force”. The capacity function of a state usually “means the state’s ability to provide its citizens with basic life chances. These include the protection from (relatively easily) avoidable harmful diseases; a basic education [...] and a basic administration that

4

This helps us to clarify our case selection: Though we are focusing on empirical statehood, we are not able to measure the empirical statehood of non-juridical states due to the lack of data. 5 For a discussion about the link between statehood and democracy see Erdmann (2014), Lauth (2004), Merkel et al. (2003, pp. 58–59) and Lauth and Kauff (2012, pp. 16–17). 6 Autocracies do not always have to be repressive. This may vary regarding the specific type of autocracy (Geddes et al. 2014; Hadenius and Teorell 2007; Kailitz and Köllner 2013; Schlenkrich and MohamadKlotzbach 2015).

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

245

regulates social and economic activities” (Grävingholt et al. 2012, p. 9). Call (2011, p. 306) defines it “as the degree that state institutions are able to provide or regulate the minimal provisions of core public goods”. Similarly, in these definitions the range of public goods which a state should deliver needs to be narrowly defined. We would stretch the concept too much if we also include broader aspects of welfare such as environmental politics and social security systems (Schneckener 2004, p. 13; Rotberg 2004b). Although these definitions address the regulatory capabilities of a state beyond that state’s task of supplying basic public goods, we are going to investigate the regulatory capabilities more closely: Regulatory capabilities allow a state to communicate with its citizens, which can be understood as a precondition of the ability to govern.7 No laws can be applied without basic administrative structures. Consequently, we call this dimension the “functioning of the administration” instead of “capacities”. These two dimensions – the functioning of a monopoly on the use of force and of the administration – are primarily rooted in a Weberian understanding of the state. Lambach et al. (2015b, pp. 1304–1305) argue that two features are characteristic of the Weberian state: first, the monopoly on the legitimate use of physical force and second, a functioning administration8. This shows the importance of including these dimensions in our definition. Although legitimacy is highlighted by Weber, we refrain from incorporating it into our definition.9 Legitimacy of the state refers to the citizens’ acceptance of “the state’s claim to be the legitimate actor to set and enforce generally binding rules” (Grävingholt et al. 2012, p. 8) or to the “social contract that binds inhabitants to an overarching polity”, as Rotberg (2004b, p. 9) puts it.10 But theoretically, legitimacy can be regarded as a promotional or obstructive factor for statehood rather than a part of the definition of statehood itself. If the state is not regarded as legitimate by the population or a part of it, it is more likely that competitors (e. g. rebel groups) will emerge to challenge the state. This adversely affects the monopoly on the use of physical force and as a result statehood is weakened. Empirically, legitimacy can be understood as a part of the political culture and is difficult to measure directly (Call 2011, p. 308).11 Therefore, indices of state fragility often include indicators of the quality of democracy (e. g. Fragile States Index) to assess it indirectly. But this results in a loss of the analytical distinction between state and regime in the measurement stage (Lauth 2004). For these theoretical and empirical reasons, we do not include legitimacy in our definition.12 7 Here also the regime type will decide, if this communication is mostly one sided/closed (autocratic regimes) or two sided/open (democratic regimes). 8 Our definition of the administration also includes some aspects of state as a service provider – a different understanding of the state (Lambach et al. 2015b). 9 In addition, we do not include the judiciary in the sense of the rule of law and political institutions in our definition as proposed by Ezrow and Frantz (2013) due to the mixing of regime and state characteristics. 10 Schneckener (2004, pp. 13–14) defines the scope of this function more widely: he additionally includes political rights or the rule of law. This is not justified because it further intertwines regime characteristics and statehood. 11 Bruce Gilley (2009, 2012) presents one possible way of measuring legitimacy on the basis of individual data, but his operationalization of state legitimacy also conflates characteristics of the state and the regime. 12

See for a deeper discussion Lambach (2008, pp. 36–42) and Marquez (2015).

K

246

O. Schlenkrich et al.

Third, we use a functional approach to measure the quality of statehood. We regard a dimension as functioning if two requirements are met: one, that the state possesses at least minimal capacities in this dimension, and, two, that these capacities are sufficient to encounter contextual challenges when they emerge.13 These contextual factors are not part of the state itself, but nevertheless they are very important for measuring statehood empirically because they represent a benchmark for the evaluation of a state’s capacities. This means statehood cannot be measured in absolute terms without regarding the contextual constraints a state faces.14 A state with minimal organizational capacities could still be a functioning state as long as it does not encounter severe contextual challenges. But the greater a threat becomes, the more organizational capacities a state has to develop to defend itself against it. On the other hand, a state with strong capacities can handle difficult contextual challenges well, whereas other states with lesser capacities could break down if they were to meet similar challenges. By capturing these interactions between organizational capacities and contextual challenges, we are able to realize and capitalize on Call’s (2011) idea of “gaps”. Thus, in order to determine the functioning of each dimension, we must additionally conceptualize the important organizational capacities and contextual challenges for each dimension. The relevant state organizations for the dimension of the monopoly on the use of physical force are the military and police capabilities15; particularly, as Hendrix has stated, the “national military is the centerpiece of the state’s repressive capabilities” (2010, p. 274; see also Tilly 1992; Spruyt 1996). With regard to the dimension of the functioning of the administration, we focus on the bureaucracy: its widespread dissemination and its quality. This is “consistent with the literature on political development, which holds that state capacity is characterized by professionalization of the state bureaucracy” (Hendrix 2010, p. 275). We also have to define contextual challenges for each dimension. Accordingly, a contextual challenge can be defined as a long-term and/or short-term factor (structural, process-related or external shock16) that a state must deal with using its capacities. Because of the multitude of possible challenges, we picked for each dimension those which are frequently described in the literature (Rotberg 2004b; Schneckener 2004): the primary challenges for the dimension of the monopoly on the use of physical force (physical challenges) are rebel groups and organized crime. In contrast,

13

It is possible that a state generates the contextual challenges due to its low organizational capacities. This could be considered as a problem of endogeneity. Nevertheless to function properly, a state has to overcome these challenges, even though it creates them. 14 See Abromeit (2004) and Stoiber (2008, 2011) who introduced a context-sensitive measurement of the quality of democracy. Grävingholt et al. (2012, p. 7) also address the context by stating that the “exact degree of authority required to maintain a stable state is dependent on context factors such as popular expectations and the strengths of rivals”. But they do not incorporate it in their overall concept. 15 It would be easy to carry the lists for each dimension forward; however, we confine ourselves to the ones which represent a quasi-consensus,. 16

See (Schneckener 2004, p. 18).

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

247

the primary challenges for the administrative dimension (administrative challenges) are natural disasters, a disadvantaged geographical location and epidemics.17 In sum, our concept of statehood focuses on the ability of state institutions to withstand contextual challenges. This differentiates our concept from other approaches in three ways: First, statehood is not just defined by the mere existence of state institutions (e. g. professionalization or number of armed forces, see Ezrow and Frantz 2013); rather the inclusion of contextual challenges allows us to assess the performance of state institutions in the light of these contextual threats. A large, well equipped army is a good sign of the existence and strength of state institutions, but we do not know yet whether they will perform well. That state could still turn out to be dysfunctional if it were confronted by an equally well-organized rebel group. Hence, contextual challenges can be seen as a “test” for state institutions. Second, we examine the functioning of the statehood in a more direct way by not relying on the outcome. As Ezrow and Frantz (2013: 1326) state: “instead of using measures of outcomes, one should actually study the state and how it functions”. Our concept emphasizes the interaction between state institutions and contextual challenges which will eventually lead to these outcomes. The outcomes, for example loss of territory or the number of battle-related deaths, which are frequently used indicators in other indices (see e. g. The Fund for Peace 2015; Grävingholt et al. 2012), are only dependent variables of the relationship between organizational capacities and contextual challenges. Finally, using a narrow definition of statehood, we differ especially from the Fragile States Index (The Fund for Peace 2015), which includes a variety of indicators for the status of the economy, democracy and the rule of law – overstretching the notion of statehood. Our concept and its implications are presented in Fig. 1: the first layer presents juridical statehood, which sets the frame for empirical statehood. The second layer is the dimension of the monopoly on the use of physical force, which is the first part of empirical statehood. It represents a necessary condition (a precondition) for the administrative dimension.18 Thus, the third layer is limited by the second layer. The contexts of each dimension are the shaded areas, which also illustrate the gaps (see also Table 1 for the CIS-concept tree and its indicators).

17

In addition, it seems that these contextual features play an important role for transformation countries – the type of countries which are mostly included in our dataset based on the Bertelsmann Transformation Index. But it would be easy to broaden the definition by including other contextual aspects as well (e. g. refugee crises, political culture) so that even non-transformation countries can be analyzed. We should be aware of a parsimonious selection of contextual factors so that a causal analysis is still possible. 18 A is a necessary condition or a precondition for B, if B cannot exist without the simultaneous existence of A. In other terms: B is a sufficient condition for A, but A can be in place without B. Thus, we argue that the monopoly on the use of physical force is only a necessary condition for the functioning of the administration – not a sufficient one. But a working administration is a sufficient condition for the monopoly on the use of force. It could be – and remains an empirical question – that the functioning of the monopoly on the use of force represents an obstacle for establishing an administration in some cases. But the critical point is that there cannot be a functioning of the administration without at least a moderate functioning of the monopoly on the use of force at the same time. The administration needs an institution which enforces its rules – and that is the monopoly on the use of force.

K

248

O. Schlenkrich et al.

Fig. 1 Layer model of the CIS-Concept. (Source: own illustration)

2.2 Measurement: indicators of the CIS

We use different sources to operationalize the bottom parts of our concept, which we discussed in the previous chapter (see Table 1): military capacity, police capacity, administrative capacity and the two contextual challenges for every dimension (rebel groups and organized crime; natural disasters and epidemics). Military capacity is measured by three indicators: military expenditure, the number of armed forces and the number of their heavy weapons. We added the aspect of ”number of heavy weapons” to the two standard indicators of military capacity (Hendrix 2010) in order to gain more accuracy by including the quality of the military equipment as well. Military expenditure is collected by the Stockholm International Peace Research Institute (SIPRI), whereas the number of armed forces is provided by the International Institute for Strategic Studies (IISS – Military Balance).19 The number of the armed forces’ heavy weapons was taken from the Global Militarization Index (GMI) by the Bonn International Center for Conversion (BICC). They use primarily the IISS’s Military Balance, but also include other sources for this indicator (Grebe 2011, p. 17). However, while these sources are quite reliable, it should be noted that the accuracy of these data may vary for some countries (SIPRI 2015b). Furthermore, to measure the military capabilities of a country, the indicators of SIPRI, IISS and BICC taken alone are rather problematic, so that “extreme caution should be exercised in drawing a link between a country’s level of military expenditure and its degree of military power or military capability, as many factors contribute to military capability” (SIPRI 2015a). But combining these different aspects into a single arithmetic mean achieves a good estimate of military strength, so that each component has equal weight and adds its part to the overall component of military capacity. 19

The data is drawn from the SIPRI Military Expenditure Database (SIPRI 2016), from the World Bank (World Bank 2016) and from GMI (2016).

K

Police Capacity Mean “Reliability of Police Service” (Global Competiveness Report) Source: World Economic Forum 2016 Scale 1–10, ES

“Confidence in Police Force; Confidence in judicial system; Have you been assaulted or mugged? Have you had money property stolen from you or another household member?” (GWP via WGI) Source: Kaufmann and Kraay 2015 Scale 1–10, Poll

Military Capacity Mean Military Expenditure per Capita (SIPRI) Source: SIPRI 2016 log10 Scale: 1–10, OD

Number of Soldiers per Capita (IISS via World Bank) Source: World Bank 2016 log10 Scale 1–10, OD “number of an armed forces’ heavy weapons in relation to the total population” (GMI) Source: GMI 2016 Scale 1–10, OD

Source: own presentation Type of data: OD objective data, ES expert survey, Poll opinion poll

Indicators

Functioning of the monopoly on the use of physical force organizational capacities contextual challenge Combined Physical Capacities Mean Conflict Intensity (BTI, 13.3) Minimum threshold: 2 Rescaling: 3 and 1 to 2, than stretching remaining values to 10 Source: Bertelsmann Stiftung 2014d Scale: 2–10, ES

Physical Challenge: Rebel Groups and Organized Crime

Government Effectiveness (WGI) Source: Kaufmann and Kraay 2015 Scale: 1–10, ES

Basic Administration (BTI, 1.4) Source: Bertelsmann Stiftung 2014d Scale: 1–10, ES

Structural Constraints (BTI, 13.1) Minimum threshold: 2 Rescaling: 3 and 1 to 2, than stretching remaining values to 10 Source: Bertelsmann Stiftung 2014d Scale: 2–10, ES

Functioning of the administration organizational capacities contextual challenge Administrative Challenge: Administrative Natural disasters, epiCapacities demics Mean

Aggregation Statehood p Monopoly of physical force * administration score of statehood cannot be higher than the value of the functioning of the monopoly on the use of physical force, if the value for the monopoly on the use of force is classified as failing (<1)

Table 1 CIS-concept tree with steps of aggregation and indicators

The contextualized index of statehood (CIS): assessing the interaction between contextual... 249

K

250

O. Schlenkrich et al.

It was not possible to acquire objective indicators for the capacity of the police force.20 Accordingly, we had to rely on survey data from the Global Competiveness Report (GCR) by the World Economic Forum and of the Gallup World Poll (GWP). The GCR arranges an annual expert survey of the business community (Executive Opinion Survey) in 150 countries. For example, in the period 2012/2013, they achieved “14,059 surveys [with] an average of 100 respondents per country” (Browne et al. 2012, p. 69). Among other questions, they asked the respondents to evaluate the reliability of police services: “To what extent can police services be relied upon to enforce law and order in your country region?” (Schwab and Salai-Martín 2014, p. 538). In contrast, the GWP is a public opinion poll. It was not possible to acquire this data directly; rather we obtained it via the source data of the Rule of Law Indicator by the Worldwide Governance Indicators (WGI). Due to the merging of the source data by WGI, this indicator contains not only the question about the “confidence in the police force”, in which we are primarily interested, but also questions about the “confidence in the judicial system” and about “being robbed or mugged”. It is not possible to disaggregate them, so only a small part of this indicator correctly fits our concept. Nevertheless, we also use the mean to aggregate police capacity because the expert survey of the GCR and the public opinion survey of the GWP can be considered as complementary, in that they can correct each other’s possible measurement errors.21 Thanks to this aggregation process, we should get a better estimate.22 The indicators of the capacity of the administration are the “Government Effectiveness” from the WGI and “Basic Administration (1.4)” from the Bertelsmann Transformation Index (BTI).23 Muno (2012, p. 108) states that although the WGI is somewhat flawed conceptually, its measurement procedure must be regarded positively due to its use of extensive sources and its overall transparency. Whereas the indicator “Basic Administration” captures the “existence and scope” (Bertelsmann Stiftung 2014b, p. 17) of the administration24, the indicator “Government Effectiveness” captures “perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies.” (Kaufmann et al. 2010, p. 4)25 Despite some overlap 20

UNODC has collected data related to the number of police personnel (criminal justice system resources), but the sample size is small and most data points are missing (especially for African countries), see http://www.unodc.org/unodc/en/data-and-analysis/statistics/crime.html. 21 We did a robustness check with the indicator V113 (Confidence in Police) of the World Value Survey Wave 6 (2010–2014). Due to the small number of countries surveyed by the WVS, the sample size consists of 44 to 46 cases. The results of the Pearson Correlation indicate a sufficient robustness (see Tab. 5 in appendix), although the Gallup World Poll performs rather poorly compared to the GCR-data. 22 The data can be obtained here: World Economic Forum (2016) and Kaufmann and Kraay (2015). 23

For a discussion of the methodology of the BTI see Lauth (2010).

24

Even though the BTI-instructions for this question exclude explicitly the evaluation of the quality of the administrative structures, some coders still include this aspect. 25 The Government Effectiveness Indicator actually covers more areas which help to measure our concept more accurately: quality of the primary education, health system, infrastructure and administration, see the detailed indicator description http://www.govindicators.org.

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

251

between those two indicators, they capture quite well our concept of administrative capacity – existence and quality of bureaucracy.26 To measure contextual challenges, we use two indicators of the criterion “Level of Difficulty” from the BTI. This criterion “reflects the observation that each country’s quality of transformation is influenced by structural conditions” (Bertelsmann Stiftung 2014a, p. 124). The contextual challenge of the monopoly on the use of physical force is represented by the indicator “Conflict Intensity (13.3): How serious are social, ethnic and religious conflicts?” (Bertelsmann Stiftung 2014b, p. 36), whereas the context of the administrative dimension is evaluated by the indicator “Structural Constraints (13.1): To what extent do structural difficulties constrain the political leadership’s governance capacity?” (Bertelsmann Stiftung 2014b, p. 35). Although the indicator “Conflict Intensity”, because of the way the question is formulated, does not catch the core idea of our concept of contextual challenge in the dimension “Functioning of the monopoly on the use of physical force”, i. e. strength of rebel groups and organized crime, many BTI-coders also take into account the impact of rebel groups and organized crime wherefore we decided to use it as the best possible solution. For example, in Senegal’s country report “[r]ebel forces have acquired new armaments and recruited militias; these groups are financed in large parts by international drug trafficking and receive silent support from neighboring Guinea-Bissau and the anti-Senegalese interests of the Gambia” (Bertelsmann Stiftung 2014c, p. 23). Regarding the indicator “Structural Constraints”, the coders are explicitly urged to consider amongst others “a disadvantageous geographical location (e. g., landlocked or small island states), [...] natural disasters [and] pandemics, such as widespread HIV/AIDS infections” (Bertelsmann Stiftung 2014b, p. 35). This fits well with our concept of contextual challenge in the dimension “Functioning of the administration”. However, there are some aspects (e. g. lack of educated labor force or severe infrastructural deficiencies) which exceed our concept. Even though these indicators are in many ways problematic, we think they are currently the best available and are sufficient for a rough estimate of the contextual challenges.27 Lastly, all indicators are transformed to a scale from 1–10 to meet the BTI-scale.28 Beforehand, the military indicators were transformed by the logarithm with the factor 10 to “increase the compatibility between different indicators and to prevent extreme values from creating distortions” (Grebe and Mutschler 2015, p. 4).29 A minimum value of 2 was set for the contextual challenges in order to fulfill the requirement that a state should possess at least minimal organizational capacities. In addition, the value 3 of the indicators for contextual challenges was set to 2 and the remaining 26

See for the BTI (Bertelsmann Stiftung 2014a) and for the WGI Kaufmann and Kraay (2015).

27

Another good indicator would be “MAGFIGHT” from the State Failure Problem Set by the Political Instability Task Force (Marshall et al. 2015). MAGFIGHT evaluates the number of rebel combatants and activists. This would be closer to our concept, but the number of coded cases is considerably smaller compared to the BTI. 28 We scaled the values twice: first for each year, and after that for all years to attain a better comparability across the years. 29

This is often done with indicators of military capacities, see Hendrix (2010).

K

252

O. Schlenkrich et al.

values were stretched to 10. This was necessary to ensure that the contextual challenges fit the first qualitative cut of the BTI-scale (which is set to 4) and thus are actually increasing. As a result of our use of the BTI-indicators the number of our cases is bound to 129 developing and transformation countries from 2006 to 2012 on a two-year basis. Our dataset contains 117 countries and 453 cases30 in total (see Table 2 the appendix for summary statistics). 2.3 Aggregation: the variable threshold

We aggregate the military and police capacities to the physical capacities of a state by calculating the mean. This means they are considered as having equal weight and compensate for one another. It is also possible to consider them as necessary conditions which would end in the use of multiplication as the aggregation rule.31 But it could be that regimes place different emphases on the component of physical capacities and assign different tasks to them, for example military regimes supporting the military more heavily or the military carrying out some functions of the police. Because of this uncertainty, the arithmetic mean seems to provide a better fit. Similarly, we proceed with the administrative capacities. Though the indicators of the BTI and WGI focus on different aspects of the administration (existence and scope of the administration, quality of bureaucracy), they also overlap. By combining them with the mean we can reproduce our concept and at the same time reduce measurement errors. To estimate the functioning of the two dimensions, we introduce a new type of threshold – the variable threshold. Methodologically, thresholds as a method of classification allow a gradual definition of an empirical phenomenon: rather than being absent, an attribute is only partially present (Lauth et al. 2014, p. 44). While we adapt the functional logic of thresholds, we relax the assumption that the cutting points should be at the same spot for every case. To achieve a contextualized estimation, the variable threshold lets the cutting points vary between groups with different contexts. This has the additional advantage that we can use the variable threshold not only as a method of classification, but also as a method of aggregation. Furthermore, the variable threshold enables us to aggregate organizational capacities

30

The following countries are missing: Bhutan (2006–2012), Cuba (2006–2012), Eritrea (2006–2012), Guinea (2008–2012), Kosovo (2008–2012), Lesotho (2008), Mauritius (2008), Myanmar (2006–2012), North Korea (2006–2012), Oman (2008–2012), Panama (2006), Papua New Guinea (2006–2012), Somalia (2006–2012), South Sudan (2012), Sudan (2010–2012), Taiwan (2006–2012), Tajikistan (2008–2012), Togo (2006), Turkmenistan (2006–2012) and Uzbekistan (2006–2012). To exclude the possibility that missing data in the CIS-dataset is correlated with increasing state fragility, we compared our missing values to two other state fragility indices which both cover more cases – the BTI Indicator “monopoly on the use of force” and the WGI index “Political Stability”. It seems that the missing data in our dataset and state fragility are only weakly correlated (see Fig. 7 in the appendix). 31 See for different aggregation rules and their implications Munck and Verkuilen (2002, p. 24): “if the aggregation of two attributes is at issue and one’s theory indicates that they both have the same weight, one would simply add the scores of both attributes. If one’s theory indicates that both attributes are necessary features, one could multiply both scores, and if one’s theory indicates that both attributers are sufficient features, one could take the score of the highest attribute”.

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

253

Fig. 2 Interaction of state capacities and contextual challenges, here dimension “Functioning of the monopoly on the use of physical force”. (Source: own illustration)

and contextual challenges to one dimension while simultaneously evaluating the functioning of this dimension. We set off the organizational capacities against the contextual challenges. Thereby, the contextual challenges serve as variable thresholds. This allows us to aggregate different values of the contextual challenges and organizational capacities to the same scale by dividing the capacities by the contextual challenge. This scale expresses the proportion of organizational capacity in relation to the contextual challenge in percentage. The values of the scale can range from >0 to 5 because the maximum value for the capacities is 10 and the minimum value of the contextual challenge is 2. Values from 0 to 1 mean that the organizational capacities are less developed in contrast to the contextual challenges (gap). We classify those dimensions as failing. Values over 1 show that the organizational capacities are more pronounced than the contextual challenges (cushion). Hence, we classify the dimension as functioning. The variable threshold makes different groups of cases comparable by acknowledging the different contexts of these groups. The variable threshold allows us to evaluate whether the capacities of states are sufficient to counter their specific contextual challenges – determining whether state capacities or state institutions can be considered as weak (gap) or strong (cushion). This relationship is shown in Fig. 2: country A’s monopoly on the use of physical force is classified as failing, even though it possesses a moderate organizational capacity (value: 5). But its capacity is not sufficient to deal with the greater contextual challenge (value: 8). Thus, the result is a gap between its organizational capacity and its contextual challenge (value: 0.63, because 5/8). In contrast, country B’s monopoly on the use of physical force would be classified as functioning. While it has the same organizational capacity as country A (value: 5), it faces only a small contextual challenge (value: 2).

K

254

O. Schlenkrich et al.

Therefore, this results in a cushion between those two variables (value: 2.5, because 5/2).32 We also intend to calculate a total score for empirical statehood by combining the two dimensions. This score reveals the conceptually proposed interaction between those two dimensions, although it necessarily goes along with a loss of information (Call 2011; Grävingholt et al. 2012; Munck and Verkuilen 2002).33 That is why we want to conserve mathematically as much information as possible: The total score for statehood is the result of the multiplication of the two dimensions because they are considered as necessary conditions in our concept. Thus, no compensation should be allowed, and low values in one of these dimensions should be penalized. Furthermore, we have to take into account that a moderate functioning of the monopoly on the use of physical force is a precondition of the functioning of the administration. Due to this, the total score cannot receive a higher value than the value of the dimension of the monopoly on the use of physical force, if the value for the monopoly on the use of force is classified as failing (values <1): Whereas a degradation from the value of the functioning of the monopoly on the use force is then possible, an enhancement is not (see Table 2 in the appendix for summary statistics).34

3 Investigating the validity of the contextualized index of statehood 3.1 Construct-related validity: exploratory factor analysis

Construct-related validity means “to investigate whether the relationship among item scores or score on parts of the test are as expected from the theory of the construct” (Algina and Penfield 2009, p. 118). An exploratory factor analysis was performed to evaluate whether the selected indicators capture our latent construct which distinguishes between military capacities, police capacities and administrative capacities. We decided not to run a principal component analysis because it is possible that the data include measurement errors. Thus, exploratory factor analysis has the advantage that it includes an additional error term for measurement errors (Wolff and Bacher 2010). Fig. 3 shows that all factors load as theoretically expected. In addition, the military, police and administrative capacities are somewhat correlated with each other. This also fits with our concept because it can be assumed that the different capacities, like the dimensions, support each other. It can be concluded that the CIS passed the test of construct validity.

32

Here is an example from our dataset: Nigeria in 2006 had a physical capacity value of 3.32 and a high contextual challenge value of 7.71, whereas Costa Rica in 2006 had similar physical capacities (3.33) but lower contextual challenges (2.00). This results in a low value for the functioning of the monopoly on the use force in Nigeria (0.53) and a rather high value for Costa Rica (1.68). 33 The more disaggregated levels (monopoly on the use of force, administration, physical capacities, administrative capacities and the physical and administrative challenges) will be available as well. 34 See the comments on the Combined Index of Democracy (CID), which uses a similar method (Lauth 2010).

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

255

Fig. 3 Results of the exploratory factor analysis (oblimin rotation). Source: own calculation with R-package “Psych” (psych_1.4.8.11). (The widely used R-package “psych” was developed at the Northwestern University [http://personality-project.org/r/ psych/].)

3.2 Criterion-related validity: the relationship between the quality of statehood and the quality of democracy

Criterion-related validity “refers to how much a predictor relates to a criterion” (Muchinski 2006, p. 95), and we test it by analyzing the relationship between the quality of statehood and the quality of democracy. Some scientists argue that states with democratic regimes are more stable in the long-run than states with autocratic regimes (e. g. Schneckener 2004), whereas others conclude that the regime distinction is in fact curvilinear and not linear. This means that democratic regimes do not necessarily show higher levels of statehood; instead high levels of statehood can be found at both ends of the regime continuum. Consequently, the objection to that argument is that states located in a regime transformation possess less statehood than stable democracies and autocracies (Esty et al. 1998; Lambach and Bethke 2012, p. 21). The illustrated connection, also designated as a J-shaped relationship, was discovered and confirmed in a few other studies dealing with state capacity and regime types (Bäck and Hadenius 2008; Fortin 2012, pp. 910–912). FortinRittberger (2014, pp. 1244–1245) states that it has become a “thriving research agenda”. If we measure statehood correctly, we should reveal this J-shaped relationship by performing an OLS-regression with the CIS-Data: whereas the quality of statehood measured by the CIS serves as the independent variable, we use the quality of democracy measured by the Combined Index of Democracy 3D (CID3D) as our dependent variable.35 The CID3D is based on a three dimensional concept of democratic quality, which includes freedom, equality and control (Lauth and Kauff 2012;

35

We use the logged GDP per capita as a control variable.

K

256

O. Schlenkrich et al.

Fig. 4 J-shaped relationship between Quality of Statehood and Quality of Democracy with 95%-confidence limits (OLS-regression). (Note: Low values of the CID3D indicate highly autocratic regimes, whereas high values represent democracies. Because of the high level of heteroscedasticity, which is also indicated by the Breusch-Pagan-Test, we used robust standard errors. R²: 0.72; Intercept: 1.75 (0.07) ***; KID3D: –0.32 (0.5) ***; KID3D*KID3D: 0.05 (0.01) ***; logged GDP/capita: 1.09 (0.09) ***; n = 428. Significance levels: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05. Source: own calculation and illustration based on the CID-Dataset (Lauth 2013) and CIS-Dataset)

Lauth 2013).36 It combines the most-used indices of democracy Polity (DEMOCscale), Freedom House (Political Rights-scale) and the WGI indicator “Rule of Law” compensating for weaknesses of these indices. This approach seems to be better than using just one of the indices by itself. Fig. 4 shows a significant curvilinear relationship: highly democratic regimes as well as highly autocratic regimes can have a high quality of statehood, whereas soft autocracies, hybrid regimes and deficient democracies have a lower quality of statehood. Because we can replicate the findings of the J-shaped relationship between the quality of statehood and quality of democracy, it can be pointed out that the CIS passes the test of criterion-related validity.

4 Empirical findings 4.1 Descriptive statistics: failing and functioning states from 2006 to 2012

For the empirical findings, we include only states with available data for all times of measurement; thus the number of states is reduced to 109 cases per year. In this way effects on the mean are minimized to an acceptable level; otherwise the mean 36

We use the CID3D instead of the CID because the CID includes a measure for the quality of statehood. We tested this relationship with the WGI democracy index (“Voice and Accountability”) as well and gained the same results.

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

257

Fig. 5 Distribution of contextual challenges and organizational capacities. (Source: Own calculations based on the CIS-dataset)

is slightly overrated because the reduction concerns states with a potentially lower assessment such as Sudan, Eritrea or Uzbekistan.37 4.1.1 Organizational capacities and contextual challenges

Most cases have to confront only a very low or fairly low level of physical challenges; merely two cases received four times the maximum value (ten), namely Afghanistan from 2006–2010 and Syria in 2012, but 15 cases can be found within the highest category (see Fig. 5).38 The distribution of cases remains fairly constant over the years, with the exception that the lowest category shrank (from 43 to 32) and the highest category has quadrupled (from two to eight). This is underlined by the trend analysis showing no change of conflict intensity for 59 states. Only four states (Sri Lanka, Chad, Kenya, Georgia) demonstrate a significant decreasing degree of conflict intensity (–2.29) and 18 states a slight decline (–1.14). On the contrary, not only the erupting violence in Syria (+5.71) and Libya (+4.57) produce dramatic shifts, but also the states of the Arab Spring and Mali exhibit a significant increase of conflict intensity (+3.43). All in all, the global average trend displays, after a short phase of containment in 2010, a rise of conflict (mean from 3.81 in 2006 to 4.25 in 2012). In contrast to conflict intensity, there is no such concentration of cases in two categories for administrative challenges, and many fewer cases with very low challenges; therefore, there is a significant frequency of fairly high or very high challenges. The distribution over the categories is constant over the years: The majority of 80 cases have an unchanged context while 15 states have to face increasing demands (+1.14). Syria and Tunisia have to face rising constraints (+2.29). Only the Republic of 37

For more comfortable reading, some terms will be used synonymously: Physical challenges or conflict intensity, and administrative challenges or structural constraints; as well as physical capacities or security apparatus. The dimension’s designations will be abbreviated by using first (functioning of the monopoly on the use of force) and second (functioning of the administration). 38 Thus, as the transformation of the indicators’ scales for the challenges has resulted in no value lower than two, the classifying threshold (<2) does not apply here; instead the category of very low challenges is exclusively filled with the measurement value two.

K

258

O. Schlenkrich et al.

the Congo shows a clear diminution (–2.29); even so eleven states could move forward with fewer constraints (–1.14). The global mean (2012: 5.53) did not change noticeably. If one context changes from one year to another, we can state that its counterpart does not change in the opposite direction (with eight exceptions). However, it does not follow that a shift of one context indicator means a shift of the other one in the same direction; if one changes, the other one often remains constant.39 Physical capacities are approximately normally distributed. The category of very low physical capacity comprises merely two cases at three times, namely Haiti (2010–2012) and Liberia in 2008; furthermore, just 27 cases achieve a very high level. But all in all there are obviously more cases with a higher rather than a lower physical capacity. The distribution over the years is stable; only the number of cases with fairly low capacities shrank. The majority of states increased their physical capacities (75) from 2006 to 2012; on top are Nepal (+1.23), Bosnia and Herzegovina (+1.28) and Niger (+1.23). Global mean increases slightly up to 5.53. The former stable autocracies of the MENA-region lost the most in terms of physical capacities, for example Libya (2.24), Tunisia (1.68) and Syria (1.38). Administrative capacities show a right-skewed distribution, which can be interpreted as a reaction of states (at least of those states capable of doing so) to address the often high levels of structural constraints. Again a very low (11), but also a fairly low (29) level of administrative capacity is infrequent, whereas many cases have high (157) and 86 even very high ones at their disposal. Consequently, this number of cases is more than three times greater than the one with a very high level of physical capacity (27). The greatest rises of administrative capacities can be noted for Zimbabwe (+1.05), Ecuador (+0.95), and rather surprisingly for Iraq (+0.75); also Afghanistan gained 0.67 points. Of course the last named states have fewer administrative capacities all in all. The global mean is back at the initial level of 6.21 after a phase of slight increase. There is no clear relationship between increasing or decreasing development of capacities in both dimensions, which could have indicated a simultaneous correlation. Although, there are a few more cases with symmetrically (187) rather than asymmetrically shaped development (128) noted at the turns of the year, the overall trend from 2006 to 2012 shows an equal distribution (54 vs 53).40 4.1.2 Dimensions and total statehood

We can find an amount of failing and weak functioning dimension values, calculated as a ratio between challenges and capacities.41 Otherwise, the number of strong 39

Perhaps the effect of one context indicator changing on its counterpart is delayed.

40

That seems contradictory and could be declared with delayed effects or a strategy of restructuring the state’s capacities. 41 The distribution of an index should cover the complete scale of possible values and should not start or end abruptly (Ziaja 2012, p. 55): The CIS achieves the first but fails partly with regard to the second requirement. Nevertheless, this is mainly caused by the sample collection which is limited to transformation states. We would face a different picture if we could include all the OECD-states.

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

259

dimension values is around the same level. However, the picture changes if we use the terms gap and cushion: Then, 238 cases enjoy at least a minimum protection by a cushion as opposed to 138 with a gap in the functioning of the monopoly on the use of physical force. It is nearly the same for the counterpart dimension: 192 cushions versus 243 gaps. The distribution varies somewhat over the years, caused mainly by the developments described in the section above. The global mean indicates a functioning monopoly on the use of physical force (2012: 1.71) like the slightly lower functioning of the administration (2012: 1.60). With regard to the overall trends (from 2006 to 2012) for the first dimension, the states of the Arabic Awakening (e. g. Tunisia with –2.64) lead the list of the worst performers. In fact, somewhat surprisingly, Bulgaria (–1.05) and Latvia (–0.73) are among the worst ten, whereas negative trends for administration are less dramatic (e. g. Tunisia with –1.00). Positive developments in the monopoly on the use of physical force are fewer than negative developments, but some states from different world regions can be named: On top are Serbia and Vietnam (both +1.10) and also African states like Namibia (+0.59) and Niger (0.49). The administration dimension again shows smoother changes: Macedonia receives maximum plus (+0.37); that country is also notable for its contradictory development, because its monopoly on the use of physical force shrank –0.49 points. Although there is a loss in information, both dimensions are aggregated into a single value in order to provide an overview of all the cases.42 The aggregation method used is unable to display the asymmetrical shaping across both dimensions; but otherwise it reasonably indicates the general condition because the aggregation procedure via multiplication is sensitive to the weaknesses of the dimension statehood by reducing the ability to compensate values. For example: Belarus (2008) and Venezuela (2006) are direct neighbors, considering their total statehood score (1.841 vs. 1.836). But Belarus has a very strong monopoly on the use of force (3.14), whereas its administration is nearly failing (1.08). In contrast, Venezuela’s monopoly on the use of force is limited due to its low physical capacities (1.23) while its administration is strong (2.75). We can say that both states have vulnerable areas; but in order to determine which state is more deficient, case studies are necessary and would have to include the contextual challenges of each dimension. 4.2 Two-dimensional types of statehood: model-based clustering

To assess which groups or types of statehood can be identified within our dataset, we used model-based cluster analysis following the idea of Grävingholt et al. (2012). Therefore, it is “assumed that the data are generated by a mixture of underlying probability distributions in which each component represents a different group or cluster” (Fraley and Raftery 1998, p. 580). Here, we use symmetric normal distribution43, so that “the algorithm tries to fit two or more (multivariate) normal distributions within 42

In addition, we have to mention that the strict hierarchy of aggregated total statehood scores match the hierarchy of the cluster membership in all but 78 cases. 43 It is also possible to perform model-based clustering with non-normal mixture distributions (Lee and McLachlan 2013), and it is not easy to decide which one should be used.

K

260

O. Schlenkrich et al.

Fig. 6 Two-dimensional constellations of state fragility. n = 453. (Source: Own calculation based on the CIS-dataset with mclust_5.1 and ggplot2_1.0.0)

the ‘observed’ distributions of the input variables” (Grävingholt et al. 2012, p. 12). This method also provides the BIC as a fit of model specification to identify the best cluster solution, and it allows us to define the volume, shape and orientation of the clusters. We used the two dimensions “functioning of the monopoly on the use of physical force” and “functioning of the administration” as our input variables.44 We modeled the cluster to vary in volume and shape, but all models should feature coordinate axes (Fraley and Raftery 2006, p. 7) to increase the chance for a solution with a small number of clusters. Although the best solution would be seven clusters according to the BIC (–1918.643), we finally decided to distinguish six clusters (BIC: –1931), so that the best Arabic states are counted within the cluster of the best European states. The resulting typology of the cluster analysis is shown in Fig. 6. In principal, we can state that the quality of the functioning of the dimensions increases from left 44 We will test an additional model-based cluster-analysis which is based on more disaggregated data such as the organizational capacities and contextual challenges in order to discover the relationship between them and gain a more detailed view.

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

261

to right. But there are some groups with a symmetrical shaping across both dimensions, whereas others demonstrate an asymmetrical shape which could easily lead to misinterpretations. Also the method used does not result in a strictly hierarchical order of clustered cases. The starting point is termed as failed states, which includes states commonly described in the literature as failed or collapsed (e. g. Haiti and DR Congo), but also some cases for which there is no sufficient consensus (e. g. Nigeria and Pakistan); for this reason we avoid the term “collapsed”. It comprises 13.25% of the cases and 22 different states with a failed or failing monopoly on the use of physical force, even though for different reasons: For example, Sudan has fairly high physical capacities (2006: 6.53) at its disposal but faces very high challenges of conflict intensity (10); while Liberia, with fairly low physical challenges to overcome (2012: 3.14), has only fairly low physical capacities (2.70) at its disposal. The functioning of the administration is at the same level, and for a majority of cases is even worse due to fewer administrative capacities in relation to overwhelming challenges (merely one third lower than eight). The failed states are symmetrical over both dimensions; no case shows a difference between the both dimensions of more than ±0.5 points. There are some cases with a lower certainty belonging to cluster one (seven of 60),45 because they share characteristics with one or more other clusters and/or are located near the border to the next cluster such as Sierra Leone, which changes its status year by year.46 In conformity with the literature (Rotberg 2004b; Schneckener 2004), the 20.31% cases (33 different states) within the next cluster are termed failing states. With regard to the monopoly on the use of physical force, no case falls below the threshold of 0.5, indicating that they are able to maintain a minimum level of security, displayed also by enormously narrowed conflict intensity. Additionally, all but three cases can meet the same minimum requirement for the functioning of the administration although challenges remain high; that is the main distinction between them and the failed states.47 Some cases (e. g. Mozambique) succeed in building small cushions concerning the security situation, but the majority stays below the threshold of one between failing and functioning. Despite this variance, the failing of the functioning of the administration is their shared characteristic with the exception of four cases. Even though six cases show greater differences between both dimensions, failing states are rather symmetrically shaped, which creates a main difference to cluster

45

As a conservative estimation, only cases below certainty values of 70% fall within a wrong group, therefore they are called uncertain cases. 46 Tab. 3 in the appendix serves as complement to Fig. 6 and provides descriptive statistics for our main indicators so that readers receive a more detailed picture of the clusters. 47 The model-based cluster analysis verifies our conceptual argument that the monopoly on the use of force is a precondition of the administration: the first two types with the lowest quality of statehood – the failed states and the failing ones – possess two symmetrically shaped dimensions: the monopoly on the use of force has roughly the same value as the administration. At the lower levels of statehood, there seems to be no empirical reason for an asymmetrical shape of these two dimensions to be possible. This is only the case for higher levels of statehood where the administration can in fact have higher values than the monopoly on the use of force (fortified administration states) – but here the monopoly on the use of force is functioning on a moderate level.

K

262

O. Schlenkrich et al.

three. There are 13 uncertain cases which are located at the lower and higher end of the cluster; of which seven are also frontier cases48: For example, Guatemala and Uganda tended to the next higher cluster in the past, but recent developments seem to confirm their belonging in this cluster. With 158 cases, the weak states constitute the largest proportion (34.88%). There is greater variance within this group, so that three subgroups can be identified: First, of the 88 significant asymmetrically shaped cases, 78 have a better monopoly on the use of force than administrative functioning. Additionally, of 51 cases with an assessment higher than 2.0 in the first dimension, 42 receive one lower than 1.5 in the opposite dimension (e. g. Albania and Mongolia), showing that they were unable to overcome their administrative challenges (mean: 6.01), which for them are much higher than physical challenges (mean: 2.45). This connection is confirmed by looking at the cases with a functioning monopoly on the use of physical force, which also suffer from high administrative challenges and are less able to establish adequate cushions (e. g. Jordan and Paraguay). Second, there are 18 cases with a dysfunction in the monopoly on the use force while their administrations are at least functioning (more than 150% covered challenges). They confront rather higher physical than administrative challenges, but often both are moderate or even fairly high, which also blocks developing statehood (Mexico and Sri Lanka). Third, a couple of cases have equally shaped dimensions and are located at the contact surface of the boxes (see Fig. 6), flagging something like the center (e. g. Namibia and Vietnam). So, despite variances in the area of security and a combination of dimensional challenges, their shared feature is the limited functioning of the administration, pointing up the distinction between them and cases within cluster four. There are 15 uncertain cases, of which six frontier cases tend to be failing because of their dysfunctions in both dimensions in relation to the rest of the cluster. Nevertheless, only one of the frontier cases (Senegal) has a lower monopoly on the use of physical force than the maximum performer from cluster two which is why the result should be interpreted as a confirmation of the first dimension as the primary function. However, there is an overlapping area between failing and weak states, the latter differing significantly due to a better functioning administration and therefore a rather contrasting asymmetry. As mentioned before, cluster four comprises exclusively cases with a strong administration (minimum: 2.04), in contrast to the weak states (maximum: 1.80). None of the 46 (10.15%) cases falls below the critical threshold between a failing and a functioning monopoly on the use of physical force; rather all but five cases show at least assessments which imply a cushion of more than 150% capacities against challenges. Moreover, the struggling cases (e. g. Bahrain and Thailand) were 48

Not only the method produces such uncertainties, also reality is hard to catch with typologies using strict thresholds, which is why such frontier cases have to be checked case by case. In the following, cases belonging with a certainty below 60% to the given cluster and one higher than 40% belonging to another one at the same time, will be so called. We will only discuss frontier cases if necessary, but in terms of transparency their appearance will be noted; additionally, due to their defined nature, they overlap the clusters. This division between uncertain and frontier cases provides information about the internal coherence of the clusters.

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

263

able to face increasing physical challenges, so that we suggest the term fortified administrative states.49 Moreover, there are no frontier and only three uncertain cases. Typical cases are Turkey, Malaysia or Serbia. Cluster five is the smallest, comprising 34 cases (7.15%). The minimum value for the functioning of the monopoly on the use of physical force is 2.67, whereas the average functioning of the administration is below the previously named 150% cushion (mean: 1.45).50 So we designate this cluster as physically pronounced states, displaying the states’ harsh emphasis on the first dimension, which is why the administration is lagging. Not only is the average administrative capacity (mean: 6.71) lower than that of the fortified administrative states (7.83), but physically pronounced states also face higher structural constraints (4.85) than they do (2.86), which will eventually demand action. Moreover physically pronounced states have an average physical capacity of 6.50, although no case is threatened; this additionally justifies the use of the term ‘emphasis’ and therefore ‘pronounced’. Looking at the boxplots, the asymmetry is self-evident; the whiskers are quite distant from one another. Typical cases are Laos, Armenia and Libya under Gaddafi. Because of their dysfunctional administrations they have some affinity with the weak sates (six uncertain cases), but they can be differentiated from them by their lower conflict intensity and their building up of massive security apparatuses.51 The last cluster comprises exclusively cases with assessments higher than 2.5 in both dimensions (with exception of the outlier Latvia); for this reason they are termed strong states. Their feature is their outstanding administration, resulting from very high administrative capacities and no serious challenges; so that 3.90 is the minimum (Tunisia in 2006). There is some variation across the monopoly on the use of physical force because some have limited physical capacities in the absence of serious conflict intensities (e. g. Lithuania). We speculate that there could be diminishing utility for a strong security apparatus, or that the strategy to manage potential conflicts involves the administrative capacities or alternatively, if necessary, their transformation. Anyway, two thirds possess at least fairly high physical capacities, namely Chile and Croatia. Way out on top with an unblemished assessment are Singapore and the Gulf States Qatar and the United Arab Emirates.

49

As we described, they are relatives of the weak states due to their shared characteristics in the first dimension but their administration are pronounced, which is also possible because the mean of administrative challenges is nearly three points lower. Thus, the cluster of fortified administrative states is only a fuzzy mirror image of the weak states whereas it is a clear one of cluster five. 50 Oman in 2006 and Tunisia in 2008 are outliers in this cluster due to high assessments of the functioning of the administration differing from the other members but that would be apply more for cluster six which has a clearly higher level. 51 Kazakhstan (2006–2008) belongs with nearly the same probability to cluster three or five and a little to cluster four which is the most uncertain case of the method’s results but tends in the following years to become a physically pronounced state.

K

264

O. Schlenkrich et al.

5 Conclusion The presented results for the measurement of the quality of statehood are satisfactory and show the plausibility of the concept which highlights the interaction between organizational capacities and contextual challenges, and illustrates the benefit of the variable threshold: First, our narrowly defined concept sidesteps the pitfall of including regime characteristics in the definition of statehood. This leads to the empirical finding that both democratic and autocratic regimes can possess a high quality of statehood. Second, in contrast to other indices which are outcome-oriented and therefore unable to distinguish between capacities and challenges, either conceptually or empirically, our concept goes one step further: it is designed not only to measure the quality of statehood but also to allow us to investigate the underlying causality of rising and declining organizational capacities and contextual challenges.52 By drawing on the interaction between organizational capacities and contextual challenges, this approach offers a new perspective on the fragility of states: It would be inadequate to evaluate the quality of a country’s statehood by considering the state’s capacities alone. Rather, a criterion is required to identify whether the capacities of the state are able to perform their functions in a sufficient way – this criterion is the contextual challenge which a state faces. The variable threshold makes it possible to evaluate whether the capacities of the states are sufficient in encountering its specific contextual challenges by representing the extent of the cushion or gap – a further development of the gap approach by Call (2011). This emphasizes the flexibility and advantage of the contextual approach. Third, model-based clustering reveals the underlying structure of our dataset and is useful in order to distinguish groups of states based on the characteristics of their dimensional performance regarding the quality of statehood. At the lower end of the statehood continuum, we find states in which both dimensions are deteriorated (“failed states” and “failing states”), whereas the middle groups show asymmetrically shaped dimensions: on the one hand, there are two clusters of states whose monopoly on the use of physical force overall functions better than their administration does (“weak states” and “physically pronounced states”). On the other hand, there is one cluster (“fortified administration states”), which is characterized by a better functioning administrative dimension in comparison to the physical dimension. Lastly, we can distinguish states in which both dimensions show a very high value (“strong states”). Additionally, the typology illustrates different possible types of decay, from abrupt collapse in spite of high security capacities (e. g. the physically pronounced states Libya and Tunisia) to a gradually downward trend (e. g. Madagascar) and wavelike ups and downs (e. g. Burundi) as well as developing states (e. g. Brazil). Our results also have implications for policy strategies which should adapt to the different configuration of capacities and challenges in a single country (Call 2011; Grävingholt et al. 2012): Are weaknesses of statehood located in the capacities or in 52

An interesting method would be configuration analyses (e. g. QCA) which are able to deal with the “equifinality” of causality in respect to the quality of statehood (Lambach et al. 2015a, p. 3).

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

265

the contextual challenges? Should we therefore aim to improve the capacities (e. g. emergency management) or try to reduce the challenges (e. g. concessions to rebel groups)? Which dimension of statehood should such policies tackle first? A longer time frame should deliver helpful insights for the evaluation of internal and external strategies, the modality of the interconnection between both dimensions, and the identification of possible development paths. Nevertheless, there are some areas in this analysis requiring improvement: First, the operationalization does not include all de-jure states; moreover the data derives from a short time span and only every second year. Also the indicators used to measure the single components of the concept are not highly accurate. However, they seem to be the best choice considering the scarcity of data for the measurement of the quality of statehood, which furthermore is supported by the high construct validity. Additionally, we intend to itemize the components and indicators in order to gain more information about the characteristics of organizational capacities and contextual challenges. And finally, information for single countries gained by quantitative measurements is always limited. Thus, case studies are necessary complements.53 Acknowledgements We would like to thank the two anonymous reviewers, Hans-Joachim Lauth, Theresa Stawski and the participants of the Panel “State Fragility in Comparative Perspective” at the Comparative Politics Section Conference 2015 of the DVPW in Hamburg for their critical and helpful remarks on previous versions of this paper.

53

Therefore, we invented a qualitative version of this concept which includes the strategies and perceptions of state actors vis-á-vis organizational capacities and contextual challenges (see Schlenkrich et al. 2016).

K

266

O. Schlenkrich et al.

Appendix

Table 2 Summary statistics of transformed indicators and aggregated components Transformed indicators

Military Expenditure Military Personnel

Min

Max

Mean

Sd

n

1.00 1.00

10.00 10.00

5.57 6.02

1.96 1.98

453 453

Heavy Weapons

1.00

10.00

6.44

2.13

453

Military Index Reliability of Police Service

1.00 1.00

10.00 10.00

6.06 4.92

1.80 1.94

453 397

Gallup World Poll Police Index

1.00 1.00

10.00 10.00

4.89 4.79

1.91 1.80

449 453

Basic Administration (BTI)

1.00

10.00

6.83

2.12

453

Government Effectiveness (WGI) Conflict Intensity (BTI)

1.00 2.00

10.00 10.00

5.68 4.01

1.64 2.04

453 453

Structural Constraints (BTI)

2.00

10.00

5.51

2.34

453

Aggregated components

Physical Capacities Administrative Capacities

1.54 1.36

9.93 10.00

5.42 6.25

1.44 1.80

453 453

Aggregated dimensions

Functioning of the monopoly on the use of physical force Functioning of the administration

0.23

4.97

1.78

1.04

453

0.14

5.00

1.64

1.35

453

–

Statehood

0.18

4.96

1.64

1.10

453

Source: Own calculation based on CIS-Dataset Acronyms: Sd Standard deviation, n number of cases

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

267

Table 3 Descriptive statistics for the two-dimensional types of statehood Cluster

1

2

3

4

5

6

Type

–

Failed states

Failing states

Weak states

Fortified administrative states

Physically pronounced states

Strong states

Number (proportion) Conflict intensity

–

60 (13.25%)

92 (20.31%)

158 (34.88%)

46 (10.15%)

34 (7.51%)

63 (13.91%)

Structural constraints

Physical capacity

Mean

7.18

5.32

3.55

3.07

2.00

2.02

Sd Min

1.74 3.14

1.04 3.14

1.14 2.00

1.09 2.00

0.14 2.00

0.15 2.00

Max

10.00

7.71

7.71

7.71

2.00

3.14

Mean Sd

8.78 1.19

7.07 1.00

5.68 0.94

2.84 0.70

4.82 0.18

2.00 0.13

Min

6.57

4.29

3.14

2.00

3.14

2.00

Max Mean

10.00 4.10

10.00 4.68

8.86 5.42

3.14 6.01

6.57 6.49

2.00 6.79

Sd Min

1.34 1.53

1.33 3.34

1.11 2.91

0.86 3.33

0.53 5.33

0.34 4.98

6.77

7.33

8.81

8.58

8.17

9.93

3.30 1.22

5.46 0.50

6.20 1.64

7.83 0.70

6.71 0.55

8.97 0.60

Max Administrative Mean capacity Sd

Monopoly on the use of physical force

Min

1.35

2.67

3.42

5.50

4.78

7.80

Max Mean

4.99 0.58

6.73 0.91

7.76 1.69

9.26 2.09

8.37 3.25

10.00 3.38

Sd Min

0.00 0.23

0.97 0.54

0.65 0.66

0.95 1.03

0.32 2.67

0.36 1.76

0.88

1.26

2.80

3.02

4.09

4.96

0.39 0.14

0.78 0.00

1.16 1.35

2.85 0.39

1.45 0.70

4.49 0.20

Max Administration Mean Sd Min

0.14

0.46

0.45

2.08

0.85

3.90

Max

0.65

1.06

1.80

4.11

2.25

5.00

Source: Own calculation based on the CIS-dataset Acronyms: Sd Standard deviation, Min Minimum value, Max Maximum Value

K

268

O. Schlenkrich et al.

Table 4 UN-world regions compared by means Challenges

Capacities

Dimensions

Statehood total

Region

Physical

Admin

Physical

Admin

MUPF

Administration

1

Eastern Africa

4.86

7.32

4.51

5.50

1.07

0.76

0.87

2

Middle Africa

5.25

7.64

4.78

4.41

1.21

0.64

0.83

3

Northern Africa

3.94

4.97

6.70

5.99

2.20

1.36

1.69

4

Southern Africa

3.33

5.24

6.21

7.40

2.18

1.44

1.72

5

Western Africa

4.42

8.03

4.25

5.00

1.17

0.66

0.84

6 7

Caribbean Central America

3.52 3.38

5.33 5.19

3.25 4.16

5.17 6.39

1.39 1.50

1.56 1.36

1.44 1.31

8

South America Central Asia

3.60

4.49

4.91

6.69

1.71

1.99

1.80

3.29

5.43

5.21

6.02

1.95

1.19

1.52

Eastern Asia Southern Asia

2.76

4.86

6.12

7.34

2.52

2.28

2.29

6.29

7.14

5.21

5.21

0.97

0.82

0.86

12

SouthEastern Asia

4.04

4.75

6.57

7.09

2.11

1.92

1.86

13

Western Asia

4.49

4.82

7.25

6.47

2.11

1.99

1.99

14

Eastern Europe

2.37

3.60

5.51

7.68

2.51

2.79

2.58

15

Northern Europe

2.10

2.00

5.98

9.00

2.91

4.50

3.60

16

Southern Europe

2.61

3.88

6.10

7.83

2.58

2.48

2.47

9 10 11

Source: Own Calculation based on the CIS-dataset Acronyms: Admin. Administrative, MUPF Monopoly on the use of physical force

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

269

Table 5 Robustness check of Police-Capacity-indicators with World Value Survey Data Police-Capacity-Indicators

WVS V113

Police Capacity (2012) Gallup World Poll (2012)

0.79 (n = 46) 0.65 (n = 46)

Reliability of Police Service (GCR) (2012)

0.76 (n = 44)

Pearson Correlation V113 (Confidence in Police) was rescaled: on the one hand the answer categories “A great deal (1)” and “Quite a lot (2)” were merged; on the other hand “Not very much (3)” and “None at all (4)” Source: CIS-dataset and WVS Wave 6 (2010–2014)

Fig. 7 Correlation between missing data and state fragility. (Source: own calculation; BTI-Dataset and CIS-Dataset. Reading support: 9 cases out of 65 are missing in the CIS-dataset at a value of 6 by the monopoly on the use of force indicator of the BTI)

References Abromeit, Heidrun. 2004. Die Messbarkeit von Demokratie: Zur Relevanz des Kontexts. Politische Vierteljahresschrift 45:73–93. Algina, James, and Randall D. Penfield. 2009. Classical test theory. In The SAGE handbook of quantitative methods in psychology, ed. Roger E. Millsap, Alberto Maydeu-Olivares, 123–157. Los Angeles, London, Washington DC: SAGE. Bäck, Hanna, and Axel Hadenius. 2008. Democracy and state capacity: exploring a J-shaped relationship. Governance: An International Journal of Policy, Administration, and Institutions 21(1):1–24. Beisheim, Marianne, and Gunnar Folke Schuppert (eds.). 2007. Staatszerfall und Governance. BadenBaden: Nomos. Bertelsmann Stiftung (ed.). 2014a. Transformation index BTI 2014. Political management in international comparison. Gütersloh: Bertelsmann. Bertelsmann Stiftung (ed.). 2014b. BTI 2014. Codebook for country assessments. Gütersloh: Bertelsmann. Bertelsmann Stiftung. 2014c. BTI 2014 – senegal country report. Gütersloh: Bertelsmann. http://www. bti-project.de/uploads/tx_itao_download/BTI_2014_Senegal.pdf. Bertelsmann Stiftung. 2014d. BTI 2006–2014 scores. http://www.bti-project.org/fileadmin/files/BTI/ Downloads/Zusaetzliche_Downloads/BTI_2006-2014_Ergebnisse.xlsx Accessed 30. Jan 2016. Bethke, Felix S. 2012. Zuverlässig invalide – Indizes zur Messung fragiler Staatlichkeit. Zeitschrift für Vergleichende Politikwissenschaft 6:19–37. Brock, Lothar, Hans-Henrik Holm, Georg Sørensen, and Michael Stohl. 2012. Fragile states. Violence and the failure of intervention. Cambridge: Polity. Browne, Ciara, Thierry Geiger, and Tania Gutknecht. 2012. The executive opinion survey: the voice of the business community. In The global competitiveness report 2012–2013. Full data edition, ed. Klaus Schwab, Xavier Sala-i-Martín, 69–78. Geneva: World Economic Forum.

K

270

O. Schlenkrich et al.

Büttner, Annette. 2004. Staatszerfall als neues Phänomen der internationalen Politik. Theoretische Kategorisierung und empirische Überprüfung. Marburg: Tectum. Call, Charles T. 2011. Beyond the ,failed state’: toward conceptual alternatives. European Journal of International Relations 17:303–326. Englehart, Neil A. 2009. State capacity, state failure, and human rights. Journal of Peace Research 46:163–180. Erdmann, Gero. 2014. Apocalyptic triad: state failure, state disintegration and state collapse: structural problems of democracy in Africa. Zeitschrift für Vergleichende Politikwissenschaft 8:215–236. Esty, Daniel C., et al. 1998. Failed states and international security: causes, prsopects, and consequences. West Lafayette: Purdue University. http://www.comm.ucsb.edu/faculty/mstohl/failed_states/1998/ papers/gurr.html. Evans, Peter B., Dietrich Rueschemeyer, and Theda Skocpol. 1985. On the road toward a more adequate understanding of the state. In Bringing the state back, ed. Peter B. Evans, Dietrich Rueschemeyer, and Theda Skocpol, 347–366. Cambridge: Cambridge University Press. Ezrow, Natasha, and Erica Frantz. 2013. Revisiting the concept of the failed state: bringing the state back in. Third World Quarterly 34(8):1323–1338. Fortin, Jessica. 2012. Democracy? The role of state capacity in postcommunist countries. Comparative Political Studies 45(7):903–930. Fortin-Rittberger. 2014. Exploring the relationship between infrastructural and coercive state capacity. Democratization 21(7):1244–1264. Fraley, Chris, and Adrian E. Raftery. 1998. How many clusters? Which clustering method? Answers via model-based cluster analysis. The Computer Journal 41(8):578–588. Fraley, Chris, and Adrian E. Raftery. 2006. MCLUST version 3: an R package for normal mixture modeling and model-based clustering. Technical report no. 504. https://www.stat.washington.edu/ raftery/Research/PDF/fraley1999.pdf Accessed 30. Jan 2016. Geddes, Barbara, Joseph Wright, and Erica Frantz. 2014. Autocratic breakdown and regime transitions: a new data set. Perspectives on Politics 12:313–331. Gilley, Bruce. 2009. The right to rule. How states win and lose legitimacy. Columbia: Columbia University Press. Gilley, Bruce. 2012. State legitimacy: an updated dataset for 52 countries. European Journal of Political Research 51:693–699. GMI. 2016. Global militarization index. http://gmi.bicc.de/index.php?page=ranking-table Accessed 03. Jun 2016. Grävingholt, Jörn, Sebastian Ziaja, and Merle Kreibaum. 2012. State fragility: towards a multi-dimensional empirical typology. DIE Discussion Paper, Vol. 3. Bonn: German Development Institute. Grebe, Jan. 2011. The Global Militarization Index (GMI). Use of the GMI for evaluating the development orientation of states and regional militarization. Occasional Paper VII. BICC. http://gmi.bicc.de/ uploads/pdf/occasional_paper_VII_02_11_eng.pdf Accessed 30. Jan 2016. Grebe, Jan, and Max M. Mutschler. 2015. Global militarization index 2015. BICC. http://gmi.bicc.de/ uploads/pdf/2015/GMI2015_EN.pdf Accessed 30. Jan 2016. Gros, Jean-Germain. 1996. Toward a taxonomy of failed states in the new world order: decaying Somalia, Liberia, Rwanda and Haiti. Third World Quarterly 17(3):455–471. Hadenius, Axel, and Jan Teorell. 2007. Pathways from authoritarianism. Journal of Democracy 18:143–156. Helman, Gerald B., and Steven R. Ratner. 1992. Saving failed states. Foreign Policy 89:3–20. Hendrix, Cullen S. 2010. Measuring state capacity: theoretical and empirical implications for the study of civil conflict. Journal of Peace Research 47:273–285. Jackson, Robert H., and Carl Rosberg. 1982. Why African weak states persist. The empirical and the juridical in statehood. World Politics 35(1):1–25. Jackson, Robert H., and Georg Sørensen. 2007. Introduction to international relations. Theories and approaches, 3rd edn., New York: Oxford University Press. Kailitz, Steffen, and Patrick Köllner (eds.). 2013. Autokratien im Vergleich. PVS Sonderheft, Vol. 47. Baden-Baden: Nomos. Kaufmann, Daniel, and Aart Kraay. 2015. The worldwide governance indicators. Aggregated indicators of governance 1996–2014. http://info.worldbank.org/governance/wgi/wgidataset.xlsx Accessed 03. Jun 2016. Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2010. The worldwide governance indicators. Methodology and analytical issues. Policy Research Working Paper, Vol. 5430. Washington D.C.: The World Bank.

K

The contextualized index of statehood (CIS): assessing the interaction between contextual...

271

Lambach, Daniel. 2008. Staatszerfall und regionale Sicherheit. Baden-Baden: Nomos. Lambach, Daniel, and Felix Bethke. 2012. Ursachen von Staatskollaps und fragiler Staatlichkeit. Eine Übersicht über den Forschungsstand. INEF-Report, Vol. 106. Duisburg: Institut für Entwicklung und Frieden, Universität Dusiburg-Essen. Lambach, Daniel, Eva Johais, and Markus Bayer. 2015a. The causes of state collapse: results from a QCA analysis. Louvain-la-Neuve: COMPASSS working paper 2015–80. http://www.compasss. org/wpseries/LambachJohaisBayer2015.pdf Accessed 30. Jan 2016. Lambach, Daniel, Eva Johais, and Markus Bayer. 2015b. Conceptualising state collapse: an institutionalist approach. Third World Quarterly 36(7):1299–1315. Lauth, Hans-Joachim. 2004. Demokratie und Demokratiemessung: Eine konzeptionelle Grundlegung für den interkulturellen Vergleich. Wiesbaden: VS. Lauth, Hans-Joachim. 2010. Möglichkeiten und Grenzen der Demokratiemessung. Zeitschrift für Staatsund Europawissenschaften 4:498–529. Lauth, Hans-Joachim. 2013. Datensatz „Kombinierter Index der Demokratie (KID), 1996–2012“. Würzburg: Institut für Politikwissenschaft und Soziologie 2013. http://www.politikwissenschaft.uniwuerzburg.de/lehrbereiche/vergleichende/forschung/%20kombinierter_index_der_demokratie_kid/ Accessed 30. Jan 2016. Lauth, Hans-Joachim. 2015. The matrix of democracy. Three-dimensional approach to measuring the quality of democracy and regime transformations. Würzburger Arbeitspapiere zur Politikwissenschaft und Sozialforschung, Vol. 6. Würzburg: Institut für Politikwissenschaft und Soziologie. Lauth, Hans-Joachim, and Oliver Kauff. 2012. Demokratiemessung: Der KID als aggregiertes Maß für die komparative Forschung. Empirische Befunde der Regimeentwicklung von 1996 bis 2010. Würzburger Arbeitspapiere zur Politikwissenschaft und Sozialforschung, Vol. 2. Würzburg: Institut für Politikwissenschaft und Soziologie. Lauth, Hans-Joachim, Gert Pickel, and Susanne Pickel. 2014. Vergleich politischer Systeme. Paderborn: Schöningh. Lee, Sharon X., and Geoffrey J. McLachlan. 2013. Model-based clustering and classification with nonnormal mixture distributions. Statistical Methods & Applications 22(4):427–454. Marquez, Xavier. 2015. The irrelevance of legitimacy. Political Studies doi:10.1111/1467-9248.12202. Marshall, Monty G., and Benjamin R. Cole. 2014. Global report 2014. Conflict, governance, and state fragility. Center for systemic peace. http://www.systemicpeace.org/vlibrary/GlobalReport2014.pdf Accessed 30 Jan 2016. Marshall, Monty G., Ted Robert Gurr, and Barbara Harff. 2015. PITF-state failure problem set. Internal wars and failures of governance, 1955–2014. Vienna: Societal-Systems Research. http://www. systemicpeace.org/inscr/PITFProbSetCodebook2014.pdf. Mata, Javier Fabra, and Sebastian Ziaja. 2009. Users’ guide on measuring fragility. Bonn Oslo: German Development Institute/United Nation Development Programme.. Merkel, Wolfgang. 2010. Systemtransformation. Eine Einführung in die Theorie und Empirie der Transformationsforschung, 2nd edn., Wiesbaden: VS. Merkel, Wolfgang, Hans-Jürgen Puhle, Aurel Croissant, Claudia Eicher, and Peter Thiery. 2003. Theorie. Defekte Demokratie, Vol. 1. Opladen: Leske + Budrich. Muchinsky, Paul M. 2006. Psychology applied to work. An introduction to industrial and organizational psychology, 8th edn., Wadsworth: Thomson. Müller, Thomas, and Susanne Pickel. 2007. Wie lässt sich Demokratie am besten messen? Zur Konzeptqualität von Demokratie-Indizes. Politische Vierteljahresschrift 48:511–539. Munck, Gerardo L., and Jay Verkuilen. 2002. Conceptualizing and measuring democracy. Evaluating alternative indices. Comparative Political Studies 35:5–34. Muno, Wolfgang. 2012. Measuring the world: an analysis of the world bank’s worldwide governance indicators. Zeitschrift für Vergleichende Politikwissenschaft 6:87–113. Pickel, Susanne, Toralf Stark, and Wiebke Breustedt. 2015. Assessing the quality of quality measures of democracy: a theoretical framework and its empirical application. European Political Science 14:496–520. Rice, Susan E., and Patrick Stewart. 2008. Index of state weakness in the developing world. Washington, D.C.: the brookings institution. http://www.brookings.edu/~/media/Research/Files/Reports/2008/2/ weak-states-index/02_weak_states_index.PDF Accessed 30. Jan 2016. Rotberg, Robert I. (ed.). 2003. State failure and state weakness in a time of terror. Cambridge, Washington, D.C: The World Peace Foundation, Brookings Institution Press. Rotberg, Robert I. (ed.). 2004a. When states fail. Causes and consequences. Princeton Oxford: Princeton University Press.

K

272

O. Schlenkrich et al.

Rotberg, Robert I. 2004b. The failure and collapse of nation-states: breakdown, prevention, and repair. In When states fail, ed. Robert I. Rotberg, 1–50. Princeton Oxford: Princeton University Press. Sartori, Giovanni. 1970. Concept misformation in comparative politics. American Political Science Review 64:1033–1053. Schlenkrich, Oliver, and Christoph Mohamad-Klotzbach. 2015. Open and closed electoral autocracies in the (semi-)periphery from 1996 to 2010: democratization and foreign aid flows. Global Humanities – Studies in Histories, Cultures, and Societies 1:57–88. Schlenkrich, Oliver, Lukas Lemm, and Christoph Mohamad-Klotzbach. 2016. State fragility in the democratic republic of the Congo 1960–2014: a new approach for assessing the quality of statehood by analysing the relationship between capacities, challenges and state actors. In The democratic republic of the Congo. Problems, progress and prospects, ed. Julien Bobineau, Philipp Gieg, 57–86. Berlin: LIT. Schneckener, Ulrich. 2004. States at Risk – Zur Analyse fragiler Staatlichkeit. In States at Risk – Fragile Staaten als Sicherheits- und Entwicklungsproblem, ed. Ulrich Schneckener, 5–27. Berlin: Stiftung Wissenschaft und Politik. Schneckener, Ulrich (ed.). 2006. Fragile Staatlichkeit. „States at Risk“ zwischen Stabilität und Scheitern. Baden-Baden: Nomos. Schwab, Klaus, and Xavier Sala-i-Martín. 2014. Technical note and sources. In The global competitiveness report 2014–2015. Full data edition, ed. Klaus Schwab, Xavier Sala-i-Martín, 537–545. Geneva: World Economic Forum. SIPRI. 2015a. Frequently asked questions: how well does military spending measure military capability? http://www.sipri.org/research/armaments/milex/copy_of_faqs Accessed 30. Jan 2016. SIPRI. 2015b. Frequently asked questions: how reliable are SIPRI military expenditure data? http://www. sipri.org/research/armaments/milex/copy_of_faqs Accessed 30. Jan 2016. SIPRI. 2016. SIPRI military expenditure database. http://www.sipri.org/research/armaments/milex/milex_ database Accessed 04. Jun 2016. Spruyt, Hendrik. 1996. The sovereign state and its competitors. An analysis of systems change. Princeton: Princeton University Press. Stoiber, Michael. 2008. Ein neues, kontextualisiertes Maß für Demokratie: Konzeptualisierung und Operationalisierung. Zeitschrift für Politikwissenschaft 18:209–231. Stoiber, Michael. 2011. Die Qualität von Demokratien im Vergleich. Zur Bedeutung des Kontextes in der empirisch vergleichenden Demokratietheorie. Baden-Baden: Nomos. Straßner, Alexander, and Margarete Klein (eds.). 2007. Wenn Staaten scheitern.Theorie und Empirie des Staatszerfalls. Wiesbaden: VS. The Fund for Peace. 2015. Fragile states index 2015. Washington, D.C.: the Fund for Peace. http://library. fundforpeace.org/library/fragilestatesindex-2015.pdf Accessed 30. Jan 2016. Tilly, Charles. 1992. Coercion, capital and European states: AD 990 – 1992. Oxford Cambridge: WileyBlackwell. Weiss, Stefani, and Joscha Schmierer (eds.). 2007. Prekäre Staatlichkeit und international Ordnung. Wiesbaden: VS. Wolff, Hans-Georg, and Johann Bacher. 2010. Hauptkomponentenanalyse und explorative Faktorenanalyse. In Handbuch der sozialwissenschaftlichen Datenanalyse, ed. Christian Wolf, Henning Best, 333–365. Wiesbaden: VS. World Bank. 2016. Armed forces personnel, total. http://data.worldbank.org/indicator/MS.MIL.TOTL.P1 Accessed 04. Jun 2016. World Economic Forum. 2016. Global competitiveness report. http://www3.weforum.org/docs/ GCR2014-15/GCI_Dataset_2006-07-2014-15.xlsx Accessed 03. Jun 2016. Zartman, I. William (ed.). 1995. Collapsed states. The disintegration and restoration of legitimate authority. Boulder London: Lynne Rienner. Ziaja, Sebastian. 2012. What do fragility indices measure? Assessing measurement procedures and statistical proximity. Zeitschrift für Vergleichende Politikwissenschaft 6:39–64.

K