Why they fail? Explanation of abortive military coups d’état in sub-Saharan Africa, 1960–2012

The third wave of democratization has washed the African continent to a prominent degree, leaving political systems with emergent electoral institutions. New civilian candidates had to compete for office, while military rulers had to pass the legitimacy test. In other words, the army was suggested to approach one of democratic states, which implies its stay under civilian control. Some of SSA countries are claimed to have relatively succeeded in rendering soldiers obedient, e.g., Ghana and South Africa. Nevertheless, it is not to say that problems of civil-military relations have ceased to exist there. As noted, armies often observe the plurality of missions, the absence of a long-run strategy or its comprehensive evaluation; besides, common difficulties are the shortage of skilled personnel and satisfactory equipment, further exacerbated by insufficient funding. Undue political influence is also widespread, which actually incensed the institution since promotions are performed according to loyalty instead of merit. All that doubtlessly constraints effectiveness of the armed forces. Moreover, troops become regularly involved in internal security concerns, including border protection, peacekeeping and overall support to the police. Yet, whereas these issues stem majorly from the government's deficiencies, the threat of military non-compliance is still present in a latent form. The questions of procurement and organization still remains in the corps' domain, which is compounded by inadequate expertise of parliamentary committees. The unpropitious result is basically the weakening of civilian oversight (especially that of legislature, whose capabilities are sometimes even not articulated), lowered accountability and an ensuing feasible challenge to authorities.

Even though democratization has sowed the seeds, it is still hard to speak about military apolitical nature. In fact, the region has encountered 67 coup instances over the years 1990-2010, with the most recent cases having happened in Mali, Guinea-Bissau, Burundi, Burkina Faso, and Zimbabwe. This evinces quite unfavourable prospects in respect to subordination of African armies to civilian control. If intervention in politics is to continue - and by this perpetrate the oscillation between civilian and military leadership, as well as between different rulers in uniform - then coup-proofing on the part of the government will be among its first considerations.

In Africa's history, there have been enough periods when coups were prevented via appropriately utilized coup-proofing strategies. From 1960 till 1982 (when the leader retired), Cameroon saw no conspiracies against autocracy of A. Ahidjo, who relied on the corps ethnically stacked even prior to decolonization. In Togo, President Eyadema kept security forces tightly to himself, staffing them with people of own ethnicity, hence protecting the regime from the army. The similar scenario was evident in Kenya under Kenyatta, where special bodies like the General Services Unit comprised primarily the ruler's ethnic group, while the regular armed forces were quite heterogeneous.

Factually, the use of these techniques does not automatically secure from attempts of subversion. However, what is of crucial significance for authorities is that coup-proofing can foil an ongoing takeover. The 1982 Kenyan coup, when a conspiracy of an Air Force officers was brutally squashed by the mentioned General Services Unit, or a suppressed rebellion in Cameroon, upon which purged members of the Guard who were co-ethnics with the former ruler tried to seize power - these and other examples clearly indicate the capacity of coup-proofed troops to rescue unfortunate leadership. It is therefore seems very important to grasp the logic of coup-proofing, i.e. what rationale conditions the government's choice and how this choice effects coup proceeding. As should be seen in the next Chapter, the general idea lies in the lowering of personnel's ability and willingness to interfere.



2. Theoretical Framework

2.1 Why do not soldiers intervene: coup-proofing

So, what can account for the fact that overthrows themselves, as well as their attempts, do not happen? A possible framework for the answer is coup-proofing, which rulers utilize to stop their armies from meddling in politics.

This set of measures is suggested by Welch's Strategy 1 applying to the military itself, rather than to the authority constrained by multiple, typical of developing states factors in laying a claim to legitimacy. Coup-proofing is therefore about achievement of governmental control that is yet not the stage of «commitment» to [civilian] rule, which Barany ranks highest in the degree of personnel's subordination and which features in the systems with popular leadership. Only through systematic application of techniques minimizing the will to interfere can legitimization entrench in the eyes of soldiers and hence secure their commitment. Unless such point is reached, the state still needs to keep the army on a short leash since its loyalty cannot be fully guaranteed. Coup-proofing deployed by the government thus includes measures toward troops' capacity and inclination to conspire.

Institutional, or structural, coup-proofing (ICP) is probably of the most sophisticated types. It draws on the notion that coordination is of seminal importance for the military, since by nature it strives to institutional cohesion. Structural coup-proofing thus creates an obstacle to a concerted action of plotters by establishing «rival organizations that check and balance each other.» Powell (2012a) summarizes ICP in three indicators, which are the total of effective ground-combat organizations (though insignificant in Powell's results); the personnel strength of paramilitary forces vis-а-vis the army that turns out to hinder both attempts and their success; finally, the counterbalancing index by Belkin and Schofer (2003, 2005) shown to deter a victory but not bids themselves. Each indicator manifests the rivalry among security organizations: the first examines military jointly with paramilitary branches, estimating the complexity of the sector and the effect which the multiplicity of actors - who are withal potential counter-actors - can have on coup proceeding; the second directly compares human resources of military and paramilitary units; whereas the last encompasses both dimensions, capturing the expansion of military branches and the relative size of paramilitary.

De Bruin provides a comprehensive account of inter-institutional competition, saying that counterpoising of military with paramilitary forces makes the latter grudge plotters a triumph, so that conspiracies founder due to the resistance by loyal paramilitary. (Yet, counterbalancing turns out not to prevent attempts themselves.) With this, the number of endeavors during the next year should increase because the same competition magnifies corps' grievances. Bцhmelt and Pilster, conversely, reveal a U-shaped relationship between ICP and a coups onset (but not their success). Risk of an attempt is therefore minimized whenever there are two equally powerful [ground-combat] organizations, in which case status-quo becomes preferred to an open confrontation. As was said earlier, however, lowering of the army's ability to coordinate and organize may be culpable for reduction of institutional effectiveness that produces unwelcome exposure to external threats.^, For avoiding such a repercussion, coup-proofing has another means in its toolkit that contrariwise aim at reducing disposition.

Instilling of the idea about the government's supremacy can be done in two ways - either via recruiting members predisposed to fidelity by nature, or through delivering stimuli which would render soldiers obedient. The former, known as communal coup-proofing, represents one of the most potent methods: here the command or troops at large become staffed with personnel of the ruing elite's origin, which can be ethnicity/tribe, or sect, or even family. In cases of a looming takeover this strategy plays on the «we-others» juxtaposition, acting as a trigger for self-protection and consequent protection of the regime. Being frequently shown on the Arab Spring example, the motif of communal ties has been also reflected in the literature on sub-Saharan states.

There are two prominent arguments on the interaction between coup risk and ethnicity issues in the African setting. The first, articulated by P. Roessler, underlines the persistence of a moral dilemma between civil war and the army's defiance. Seeking to evade an insurgency by potentially dissenting ethnic groups, authorities decide to appease them via permitting an access to state resources, i.e. military service. Still, given that anti-government sentiments usually endure, such power-sharing may well amount to self-destruction. Now, when rival groups have acquired necessary capacity to mount a coup, the situation for the regime is not better than if there would be civil war. (Possible solution is thus seen in semi-inclusion, when rivals are cleared for civil, but not military, service: then they do not threaten by a coup much while being indulged to the degree as not to rebel.) The second argument belongs to K. Harkness contending that coups are born out of ethnic exclusion - either upon decolonization, when a first ruler adopts the policy of ethnic loyalty while the corps is heterogeneous, or upon democratization, if a new leader of a different origin observes troops stacked with a formerly dominant ethnicity. Therefore, the strategy of ethnic inclusion might come to be a safety valve for the government:^, intuitively, it should abate grievances within the military, laying the prospects for the regime's long-term survival. With this strategy, however, we move towards policies other than those exploiting socially-implanted features.

Another approach to lowering the willingness utilizes the mundane characteristics of human beings such as a desire for money or cost avoidance. During financial coup-proofing, politicians get commonly involved in spoiling the army with patronage resources to co-opt it. Higher levels of military spending or funding per soldier imply that the staff becomes conciliated with raised benefits and new equipment - which decrease both the will and capacity to intervene. (We should understand that, though initially intended to lessen disposition and hence avert attempts, this type of coup-proofing impacts on their failures as well since those coup-proofed become counter-agents to plotters and weaken their ability to succeed.) Speaking of Africa in particular, Collier and Hoeffler show that increased military expenditure substantially reduces the risk of a successful ouster (although the effect is not so strong for attempts). In contrast to democracies, which at times of high coup risk deploy diversionary conflicts, autocratic leaders widely convert defense finances into spoils that placate disgruntled soldiers, hence discarding the necessity of diversion. This pattern is as well noticed in sub-Saharan Africa, where the major part of military expenditure goes to personal allowances and salaries in lieu of basic military needs. Besides direct provision of such «an advance» for loyalty, patronage can be also dispensed through an intermediary institution, specifically a ruling party. Its establishment not only alleviates the issue of political control, but, importantly, assures clients from the corps in a long-term access to state resources. The Fathers of the African Nations as well as subsequent military rulers were able to take advantage of that technique, which let them to retain power for a fair amount of time.^,

Apart from pleasing an appetite for material benefits, there is an alternative way to reduce personnel's inclination - through the system of monitoring mechanisms that raise the costs of non-compliance. Still, this category pertains mostly to developed countries, where a capable government can assume competences of supervision and punishment.

Having indicated the principles of coup-proofing, we now proceed to describing the relationship between the government and the armed forces as it is proposed by this work.

2.2 The logic of civil-military relations

The argument proposed is based on the understanding of civil-military relations as an endogenous process. That is, the preceding actions and structure of the armed forces determine a coup-proofing strategy of the government, which in turn conditions soldiers' behaviour toward the latter. There are several Assumptions we make to delineate the logic:

1) A new government, except for the very post-independence one, comes with each successful takeover and is constituted of coup-makers or their collaborators. (I.e. besides military leadership, it can be a transitional government or elected authority so long as personnel returns to the barracks.) If there have been no successful coups previously, we surmise that the government under consideration is a continuation of the regime installed right after decolonization;

2) The army tends to reflect ethnic composition of any current government, which means that alteration in the ruling elite should affect troops' structure (though there are always representatives of other ethnicities as well).

The setup for development of civil-military relations in a country is as follows. Once the government assumes power via (in italics further), it has to determine a strategy toward the armed forces. With this, the government possesses certain knowledge upon which it bases own decision. The knowledge is limited to the track of prior coups as well as evidence of present army composition. The latter, called , indicates whether the introduced government matches troops ethnically - if yes, which implies that no ethnic change has occurred within the elite, then the leadership has little to fear; otherwise it needs to safeguard somehow from potentially menacing corps. The track of coups is constituted by two values evincing a) the prospects of a coup trap and b) the extent of general threat arising from the military. The first, denoted as , is composed of two terms, which are the score of attempts staged within the last 5 years, excluding itself, and the inverse of the number of years since an attempt preceding , i.e. . The second, denoted as , amounts to the overall percentage of success, i.e. the relationship of victorious undertakings, including , to all bids made since independence.

Thus, having ascended to power, the government is faced with a set of characteristics which influence a forthcoming decision. There are two ensuing scenarios, each encompassing a coup-proofing strategy, by which civil-military relations can develop.

1. When it is (1) a power transfer that observes ethnic change in the ruling elite (especially on those previously powerless or discriminated), which means growing ; or (2) a successful coup witnessing growing , which signifies invigorated subversion activity, and either present or absent alteration of the ruling ethnicity - then the newly-minted authority is confronted with the imminent threat of a takeover (due to either an unwelcome ethnicity or ongoing instability trend). This implies that the government needs to address the issue in the most outright and surest manner, only to retain power for the near future. The task thus comes down to effective securing of troops' loyalty, which commonly amounts to staffing them with personnel predisposed to fidelity, i.e. ethnically same people - coupled or not by pleasing troops (given that the majority is stacked) by increased allowances. At the same time, still, this reversely increases motives to intervene of other groups within the army, those left oppressed (purged, demoted) and disgruntled. Subversion is therefore very likely to originate from them. Yet, such a side effect is taken into account by the regime's strategy as ethnic stacking allows for protection from these looming countercoups. By minimizing disposition of a strictly defined category of soldiers the leadership acquires a personal guardian aimed to resist emergent attempts.^,

However, such (indeed powerful) counter-agent as allegiant ethnic corps may sometimes fail to impede conspirators' ability. In this case the latter possess not only motivation - because many of opponent factions do, from those purged hard to those biding aside - but also veritably good coordination. It hence should be a group that has sufficient resources to displace the government (in contrast to ethnicities so harassed as to be incapable of organizing), which may well be accrued through a prior growth in funding.

Strategy 1 can be therefore outlined as: vdisposition (for a certain group, i.e. others' ^) => vability of conspirators (still, there are many willing factions, and some can obtain necessary resources to coordinate efficiently). Hypothesis 1 is that given conditions (1) and (2), attempts are foiled by ethnically stacked troops, though a successful conspiracy is to be expected through the mechanism described above.

2. When it is (1) a power transfer not observing ethnic alteration in the ruling elite, i.e. dominant group(s) keep(s) in office, and noticing that is not so high to be alarming - then the government does not perceive an impending threat to own position. Feeling rather secure in the given moment, it therefore discards first-hand solutions (e.g., ethnic stacking) from consideration and instead may want to address the military's far-reaching disposition and capacity to interfere, (). That is, if authorities do care about prospective tenacity, they will most likely strive to invest in long-term survival. At first place the government aims to counteract ability of existing coupmongers who presents a menace within an indefinite term. For that purpose, structural coup-proofing appears to be a convenient accessible tool, which leads to deployment of various counterbalancing measures. (Though usually, paratroopers are deliberately bound to the regime by their origin.) However, still lasting inclination of the army to step in politics necessitates officials to handle it as well. Here, establishment of a party securing state patronage and provision of long-term economic benefits - all seeks to co-opt the military institution, thus minimizing the risk it poses. These disposition-reducing techniques will yield another loyal personnel to oppose unfolding intrigues. (Yet in a broader perspective, the strategy is intended to extensively diminish motivation to conspire for all members of the armed forces.)

(However, it is not to say that spoiling with financial resources pertains only to that coup-proofing strategy. There is possibly a need to differentiate between an immediate increase in funding, used to gratify «own people» particularly, and systematic cooptation of the army through, for instance, granting it a sizeable share in the economy.)

Strategy 2 can be outlined as: vability (directly addressed), and vdisposition (aimed at all members) => considerable vability. Hypothesis 2 is that given condition (1), attempts are systemically foiled through structural and long-oriented financial coup-proofing (plus the ruling party), purposed to constrain the capable military. In this case, a successful countercoup is executed by the tenacious armed forces still having the capacity to organize despite multiple measures enacted by the government.

With each subsequent foiled coup, the strategy adopted earlier is supposed to be enhanced.



3. Data and Model

3.1 Independent variables

The key indicators derived form the theory are the following:

- the imminent threat to perpetrators of the successful , i.e. the new government . Constituted of two terms: , the score of attempts staged within previous 5 years (without the coup), and equalling , where is the number of years since a last attempt before the coup. Each component is multiplied by weights if those bids within 5 years have been victorious, implying that then the effect of should be stronger. The weights are as follows: with one successful coup, it is 1.5; with two - 2; with three - 2.5; with more than three - 3, respectively.

- the state of the military under . Computed as the percentage of successful undertakings (with ) from all attempts since independence. The data on coups is compiled from McGowan (2003), Marshall and Marshall's (2014) Coups d'Etat dataset, Roessler (2016) and Luttwak (2016). The multiplicity of sources is needed since the coups count comprises a vital part of the theoretical framework, coupled by the restriction of data to one continent.

- the menace to as a ruling elite from other ethnic groups. Amounts to (for the year of overthrow) if is staged by powerless/discriminated or irrelevant ethnic groups; afterwards, if a foiled countercoup against occurs within assumed 5 years, the threat gets additional score 0.4 (for the year of an attempt), if two attempts in the same year - 0.5. If originates from junior/senior partner group, the threat equals ; then a failed countercoup happening within 5 years produces extra threat in 0.2, while two such bids in one year leave 0.3. If 's ethnic group is categorized as dominant (sometimes it can also be senior partner, so that we need to cross-check whether a change in the ruling elite took place), the threat receives 0. The data on conspirators' ethnicity comes from Roessler (2016), who himself borrows the status of ethnic groups from the EPR Dataset. The latter does not cover Burkina Faso, Cape Verde, Comoros, Djibouti, Equatorial Guinea, Lesotho, Mauritius, Sгo Tomй and Prнncipe, Seychelles, Somalia, Swaziland, and Tanzania either because the size criteria are not satisfied or because ethnicity does not present much concern. The scholar also omits Botswana for the second reason. This compels us to put those countries zeros for .

Given these measures for each country-year, which may encompass a coup year, it is expected that bases own decision regarding the coup-proofing strategy on these indicators. Therefore, in order to gauge the potential of the present , i.e. the prospects of its triumph in the face of concurrent coup-proofing, we need to look back at the estimates characterizing the prior and influencing 's choice. Consequently, for each attempt (successful or not as the result is yet unknown) we specify , and of the preceding overthrow. Additionally, we incorporate measures used as controls for : they are and equalling the sum of and values of each (failed) bid respectively for the time of 's rule (i.e. from the very year of the coup to the year before ); and the overall percentage of success upon , , that basically amounts to of the previous undertaking. Also, in order to account for Hypothesis 2's preconditions, we create two binary variables signifying low values of and respectively.

Since there is no systematic data on ethnic composition of African armies for the period examined, we try to simulate stacking through ethnic heterogeneity also appearing in Collier and Hoeffler (2005, 2007) and Singh (2014). Hence, we introduce Alesina et al.'s (2003) measure of ethnic fractionalisation from QoG Data Archive, surmising that higher fractionalisation evinces a greater possibility for authorities to rely on ethnic stacking. (However, we dub the variable itself as «ethnic stacking».)

To capture structural coup-proofing we deploy Bцhmelt and Pilster's (2011, 2015) number of effective ground-combat organizations, which appears to be an apt measure for our purpose. (Though we admit that it proves insignificant for coup outcomes in the papers of Bцhmelt and Pilster (2011) and Powell (2012a).) Given the amount of missing values in Effective Number, we add the minimal score if a country has values in consequent years and the first of them is less than or equal to , so that it does not exceed much the minimal score. (Herewith we presuppose that coup-proofing tends to increase gradually rather than leap.) Though we also introduce Belkin and Schofer's (2003, 2005) counterbalancing index as well as the relative strength of paramilitary to military forces featuring in Belkin and Schofer (2003, 2005) and De Bruin (2018), their effect becomes difficult to assess due to little coverage for the data.

Financial coup-proofing is reflected in accordance with Powell (2012a), Bцhmelt and Pilster (2015) and De Bruin (2018) as the change in military expenditure and logged military expenditure to military personnel taken from the Correlates of War project. Both might well evince the amount of granted resources aimed at securing soldiers' loyalty. Plus we incorporate military expenditure as a percentage of GDP used by Collier and Hoeffler (2005, 2007), which is anticipated to assuage the corps' interests in the long run. ^, (Unfortunately, we do not have access to data mentioned in Albrecht (2015a), e.g., on military economic enclaves or individual entrepreneurship, that would allow to perform deeper analysis.)

The presence of the ruling party is also taken into account via a binary variable, where 1 is put if an evident dominant party exists, and 0 if it is not the case. Particularly, it does not mean that the regime is necessarily authoritarian: even if free and fair elections do take place, one party may still hold the major share of power, as in the case of Botswana Democratic Party today. The data has been collected manually.

3.2 Controls

During analysis we control for additional factors which might impact on coup occurrence. Thus, poverty and slow economic growth are shown to endanger political status-quo, making elites vulnerable to forced removals. We therefore include economic indicators - logged real GDP per capita and change in real GDP per capita - based on Gleditsch's (2002) Expanded GDP data (Version 6.0).^,

Regime variable is considered critical with respect to subversion activity, although interpretations given by authors may differ. On the one hand, democracies are liable to fewer takeovers because of greater civilian control and, notably, popular legitimacy. However, while willingness to interfere may indeed subside, the chances of success actually increase as the government becomes less oppressive. On the other hand, military regimes - particularly collegial military rule - are regarded as driving both attempts and their victory, given the propensity for prompting countercoups. Hence, two separate controls are incorporated as done by Powell (2012a) and Bцhmelt and Pilster (2015): for regime type - authoritarian or democratic - and military regime in particular. The latter indicates whether a coup is staged against an acting military government, taken from McGowan (2003). The former, drawn from Polity IV index (Polity2 score; 2018 update), should address the difference in vulnerability to ouster in opposite political settings. Given the influence major powers exerted on dependent states during the Cold War, this period is also taken into account via a binary variable getting 1 until the year 1991; zero value, by default, denotes the subsequent era of democratization in Africa.

The presence of political instability is widely accepted as a critical factor for government overthrow. This leads us to include the measure of instability consistently with Singh (2014), who derives it from the PRIO Armed Conflict Dataset (Version 3-2005). The civil conflict variable is binary, where 1 corresponds to the highest values of 2 or 3 in the internal armed conflict and the internationalized internal armed conflict indicators (Type 3 and Type 4 respectively) of the PRIO Dataset. Additionally, the international conflict control is equivalent to the same values in the interstate armed conflict indicator (Type 2).^,

The number of military personnel also appears to be a determinant: the bigger the military's size, the harder to undertake seizure due to the lack of coordination and heightened risk of resistance by others. To capture this possible handicap, the logged number of personnel lagged by one year is introduced. However, we include this as control rather than a core variable because it can hardly be seen in practice that the government purposefully increases own army as a part of counterbalancing.

One of the factors regarded as strongly contributing to a coup's outcome is the rank of perpetrators. Actually, where subversion originates from not only reflects the rationale for scheming but also defines potential dangers plotters may encounter. Generally speaking, the higher the position of coup makers, the greater the chances of success as other unresolved personnel become convinced of victory more securely. However, an upper rank is not always a clue to triumph - what matters is how effectively the phase of «making a fact» has been accomplished. We therefore follow De Bruin (2018), coding attempts staged by at least one general (that is, military ranks with «general» in the name) as coups from the top, those led by majors, colonels or brigadiers (as well as their subtypes) as coups from the middle, and all the rest as coups from the bottom.^,

Finally, there is a binary variable signifying whether it is a first attempt for a country's coup history. This stems from the discussion on dictatorships founded by Black leaders right after decolonisation and the military deemed a saviour from those authoritarian trends. Hence, an intuitive guess is that first instances of such power transfer should be victorious.

3.3 Model

The next part aims to test Hypotheses 1 and 2 using time-series cross-section data on sub-Saharan Africa over the years 1960-2012. Although this indeed implies the check of whether some conditions take place to prompt a specific outcome, which may allude to QCA analysis - we nonetheless stay with logistic regression with an interaction effect. It is primarily due to the number of dimensions that our cases vary by as well as the type of raw data, which is real quantitative data suitable for conventional statistics. Interaction is supposed to produce the same effect as QCA, not missing the core idea.

The logistic model is organized such as to reflect the hypotheses deduced from the theory. Our dependent variable is an outcome of a staged attempt, where successful takeovers receive 1 - which we test against all existing attempts. We do not test it against all the years with and without attempts since we deem it necessary for an undertaking to emerge.

The key independent variables described above - ethnic stacking, change in military expenditure, logged military expenditure per soldier, military spending as a GDP percentage, number of effective ground-combat organizations, and ruling party - are subject to interaction with , (first three) and (last three) characterising the antecedent , as well as prerequisites for Strategy 2 - and (last three). In one sequence of brackets we introduce Hypothesis 1, i.e. the three first predictors together with and , while in another sequence Hypothesis 2, which is the rest of predictors together with , and . The logic here is that coup-proofing techniques should be combined with indicators leading to them. and are plus-separated - being «OR» conditions according to the theory - whereas , and become multiplied since they are «AND» conditions.

Besides, there are certain control estimates added to main moderators. The percentage of preceding success is intuitively joined to in the expectation that its increase should contribute to a positive outcome. measures, for both and , by definition stand for the situation when some score of prior abortive undertakings is «accumulated» over 's rule. Therefore, they are specifically used to gauge the influence of coup-proofing from the long-term perspective. We set control for Hypothesis 1 and for Hypothesis 2: while fits the logic only of Hypothesis 1, seems apt to denote the period with effectively prevented takeovers in the second case.

All the other factors listed in the section «Controls» are placed after the brackets as in an additive model.

Furthermore, some variables observe a high correlation between them (with ), meaning that they cannot be run in one model simultaneously. They are pairs -, -, -, -, -. We therefore include each from a pair by one in a new model, so as to avoid the problem of multicollinearity.



4. Analysis

In analysis we run 12 regression models as outlined below, searching for common patterns revealed by them.

First we introduce full interaction for Hypothesis 1, which is three parts of coup-proofing Strategy 1 multiplied by two moderators - whereas Hypothesis 2's interaction is left unfulfilled, i.e. without the prerequisites for initiation of Strategy 2. The output can be seen in Table 1. Speaking of results by single independent variable, either predictor and moderator, a positive effect of is shown to be significant at the 0.05 level in 4 out of 4 possible models. Even if standing outside of combinations with coup-proofing, which is needed to prove or disprove the effectiveness of policies handling the imminent threat - a positive sign is quite expected. High well suggests the period of [army-driven] political instability that makes a country see new government after another. Such phenomenon has been widely recognised in the literature, being usually called a coup trap. The percentage of victorious takeovers preceding a bid, , should also contribute to a successful outcome as strongly, having significance in each of 2 models where it appears. This well describes the situation when the army knows the state of art and possesses a «knack» for staging coups. Benin, prior to resolute power seizure by Mathieu Kerekou in 1972 and subsequent establishment of personal dictatorship, is pertinently said to possess «coup-mania» as the country witnessed five military overthrows (and none failed) from 1963 to 1969. An excellent example here can be Burkina Faso where all six attempts over the years 1966-1987 succeeded in removing authorities.

The number of effective ground-combat organizations, which refers to coup-proofing techniques, possess a positive coefficient (significant at 10% in the models with instead of ) if tested without interaction. That might seem strange at first sight since the direction anticipated is inverse. However, our finding can be corroborated by Bцhmelt and Pilster's (2015) analysis that discovers a U-shaped relationship between the square term of Effective Number and coup occurrence (though, the authors confirm it only for attempts). Indeed, if we introduce the square measure in the model, this variable becomes significant at the same level and with the same sign - i.e. the probability of success is minimized at the point of approximately two rival organizations, after which it starts to grow. Whereas such parabolic trend is not confirmed by the scholars within a global sample, it turns out to find evidence in the data on Africa.

Table 1. Regression Results

	Dependent variable:
	Successful coup
	(1)	(2)	(3)	(4)
a	20.554**	22.689**	20.564**	22.479**
	(8.984)	(9.271)	(9.136)	(9.523)
Effective Number	0.605	0.607	1.647*	1.677*
	(0.835)	(0.842)	(0.964)	(0.966)
bt-1			4.604**	4.940**
			(2.181)	(2.213)
Military regime	-1.728***	-1.691***	-1.589**	-1.560**
	(0.644)	(0.634)	(0.652)	(0.643)
GDP pc change	-4.828*	-4.563	-4.044	-3.687
	(2.774)	(2.798)	(2.773)	(2.799)
GDP pc (log)	-0.650	-0.566	-0.975*	-0.916*
	(0.524)	(0.523)	(0.540)	(0.539)
Civil conflict	0.813	0.898	0.966*	1.052*
	(0.572)	(0.574)	(0.580)	(0.585)
Cold War	1.842**	1.779**	1.799**	1.738**
	(0.769)	(0.770)	(0.776)	(0.782)
From top	2.539***	2.462***	2.451***	2.360***
	(0.654)	(0.654)	(0.672)	(0.673)
Expenditure per soldier: a	-2.788**	-3.092**	-2.779**	-3.056**
	(1.197)	(1.240)	(1.211)	(1.267)
Ethnic stacking: a	-23.283**	-26.069**	-23.633**	-26.051**
	(11.003)	(11.469)	(11.184)	(11.772)
Effective Number: bt-1			-3.173*	-3.399**
			(1.622)	(1.647)
Expenditure as GDP%: bt-1			1.460*	1.478*
			(0.813)	(0.836)
Expenditure per soldier: Ethnic stacking: a	3.147**	3.550**	3.173**	3.533**
	(1.455)	(1.524)	(1.470)	(1.556)
Constant	-0.427	-0.631	0.361	0.305
	(3.890)	(3.865)	(3.931)	(3.914)
Observations	172	172	172	172
Log Likelihood	-79.840	-80.616	-77.520	-77.979
Akaike Inf. Crit.	235.681	237.232	231.041	231.958
	*p<0.1; **p<0.05; ***p<0.01

Note. Robust standard errors in parentheses. Only significant variables are presented. For the full table - see Appendix.

The interactions part starts with a significant double term involving moderator and funding per soldier. The predictor exerts a reverse impact on an outcome with significance at the 0.05 level. Although we have not made a suggestion on this particular double interaction (since ethnic stacking has been set as a necessary component of Strategy 1), the result can still be read in terms of Hypothesis 1. Leaders being at a risky position are prone to reduce willingness of the military by granting it «toys» and raised benefits, so that those coup proofed parts present a counterweight to subversion initiatives. This completely agrees with the thesis on buying loyalty through economic incentives. In particular, Powell (2012a, 2012b) notes that such a story applies a lot to African governments spending an inordinate portion of money on wages and personal allowances.

Further, in all the four models, ethnically stacked troops proxied with ethnic fractionalization are shown to be accompanied by high imminent threat upon a last victorious coup. That is, the situation of political turmoil [caused by the military] makes a newly-minted leadership feel unsecure, inducing it to adopt any measures which would guarantee «political tomorrow». According to our output, ethnic stacking can represent a good solution to the problem: the predictor has a negative coefficient statistically significant at the 0.05 level. Since the discussed interaction term considers only , supporting cases should have substantial values in this estimate concurrently with low or moderate . The 1981 Ghanaian coup that ushered Jerry Rawlings in politics for the next two decades is marked by , which is a very high score. Indeed, the country experienced six another coups since Nkrumah's removal in 1966, with three of them successful. During subsequent years Rawlings's regime is said to promote the position of the Ewe group that is basically his own ethnicity. Togolese President Lt. Col. G. Eyadema, who assumed power through a countercoup in 1967, was largely backed by Kabaye people prevailing «the army, the police and the bureaucracy». Though both groups the leaders came from were a «junior partner» upon the takeover - implying that the threat in terms of ethnicity was not of primary concern - representatives of either began to control sensitive state spheres under new authorities. Another apt example is provided by Idi Amin in Uganda, commonly infamous for his ethnic purges in the military. In fact, the future dictator originated not even from a «partner» ethnicity but from a dominant one; albeit, he indeed ascended to office at the moment with two conspiracies (attempted and successful) staged recently. All the three examples manifest avoidance of ensuing ousters [right] after the seizure: Ghana saw no removals since then, Togo witnessed it only in 2005, whereas Uganda's next overthrow was in 1979. Meanwhile, the number of thwarted attempts in the interim equalled 3 under Rawlings, 2 under Eyadema and unprecedented 8 under Amin. We can thus infer that the securing effect of combined and stacking asserted by Hypothesis 1 is upheld.

It is crucial to point out that the afore-mentioned interaction does not encompass supplementary indicators of financial rewards. Once they are included, however, the result turns out to be quite opposite. If accompanied by per soldier funding, the previous term starts to exert a positive influence on the likelihood of success, with dependence significant at 5% in each model. This actually connotes a curious idea: increased resources in combination with ethnic stacking (both stemming from high ) tend to induce ultimate subversion, rather than when either technique is applied separately. The logic behind it could be that a rise in financing leads to diffusion of resources and their [unintended] accumulation by potential conspirators. Stacked troops then may presumably become a trigger prompting disgruntled parts to mobilize available means. Because some have finally reached enough organizational capacities, plotters get able to surmount allegiant personnel in staging a takeover.

The triple interaction term with the change in military spending, on the contrary, has a negative coefficient not getting significance. (Though, the very sign alludes to the previously discussed patterns where the coup likelihood is diminished. Besides, such distribution of the effect between the two economic variables is quite surprising, as spoils have been deemed by us to be more targeted and therefore kept away from distrusted contingents.) On account of the fact that only a positive impact is statistically corroborated, we should make a discouraging conclusion, politically speaking, about individualised rewards. Money become more a poison than a cure for authorities' survival when they are merged with ethnic stacking. This substantiates our thesis on Strategy 1's feasible precariousness whenever rival groups manage to get empowered.

Although the first section primarily aims to check Hypothesis 1, coup-proofing tools from Strategy 2 also gain statistical significance in the two models involving in lieu of . Thus, coupled by moderator , predictor Effective Number observes a negative coefficient with p-value equalling 0.05 and 0.043. This output bears two potential implications regarding our expectations in Hypothesis 2. On the one hand, the link between overall performance of the military in coup activity and counterbalancing techniques devised to hamper it appears to be corroborated. That is, a higher percentage of success antecedent to an attempt prompts leaders to utilize measures reducing soldiers' ability to seize power. On the other hand, however, the term features specifically but not that has been supposed by us to be a key factor for embarking on structural coup-proofing. As can be seen, Effective Number gets the same sign in combination with , yet fails to obtain statistical evidence. Such finding suggests that it should not be the very beginning of rule which accounts for counterbalancing: its application [and influence] is better reflected through the percentage of success closer to an endeavor. (Technically speaking, such an effect is created through existence of rule spans in which the score of vying organizations rises somewhere in the midst of abortive countercoups - e.g., in CAR following the 1981 ouster led by Gen. Andrй Kolingba, after the 1984 Guinean takeover staged by Col. Lansana Contй, or in the wake of Flight Lt. J. Rawlings's interference in Ghanaian politics.)

Strategy 2's another component is granting the army a stake in the economy, which we try to capture via military expenditure as a percentage of GDP. The variable is significant at 10% - but the positive sign is directly the opposite of what has been anticipated. Rather than lessening the outcome probability, this factor turns out to generate consequent subversion given the rise in preceding success. The reason for that then should be akin to the unwitting effect of financial rewards: money inflows may bear seeds of demise for a government making them. Similarly, Albrecht (2015a) argues that there are two different approaches to economic coup-proofing, and one of them eventually brings about an actor in its own right with accumulated [and dispersed] resources. Since here we do not multiply Hypothesis 2' variables by the required preconditions, significance of the result is most likely inflated by all those cases featuring coup after coup, where went up as often concurrently did a GDP percentage. But it is basically the scenario of Hypothesis 1, and therefore should not be referred to the analyzed coup-proofing tool. (Hence we should look at its performance in the section that follows.)

Now we proceed to the second part by introducing full interaction for Hypothesis 2, i.e. three components of coup-proofing Strategy 2 multiplied by two moderators and then by the two prerequisites - whereas Hypothesis 1's interaction is left unfinished. The output is provided in Table 2. In the first two models, the output by single variable shows that ethnic stacking lessens the feasibility of future overturning at the 0.1 level. It majorly agrees with the prior conclusions about the impact of securing loyalty through natural propensities. Given that no combination emerge as significant when both preconditions are run together, we therefore incorporate them by one. As a result, we witness six triple terms involving a dominant party - with and or with and either or - which draw support at 5% and 10% respectively. The revealed dependence connotes that the presence of a ruling party actually contributes to coup success. This, however, contradicts our supposition about a protecting effect of the co-opting institution in the long run. Probably, the situation should be read from another perspective that we have done: for instance, it may well be a military regime founding own party in the wake of a coup (which besides features minor or ) - but being eventually unseated after a while. In other words, establishment of a party appears to not save from toppling if soldiers are persistent in their dislike of authorities or generally know how to stage takeovers. Aside from spurious relationship that might be the case if association is coincidental, a dominant party in such settings can serve as an additional venue for officers to organize subversion. A good illustration of the first type is the ultimate deposition of Sudan's President Jaafar Nimeiry in 1985, who himself seized leadership sixteen years earlier and, though having the Sudanese Socialist Union since 1971, survived six coups in the interim. As for the second type, Jean-Bйdel Bokassa's rule in Central African Republic demonstrates it aptly: his bid for power in 1966 was first for the country's history, and owning of MESAN did not safeguard the emperor from own military in 1979.

Although such conclusion might seem unobvious, the output compels us to think consistently with the logic just described. Indeed, in other combinations rather than delineated in the foregoing paragraph, the party variable is characterised by negative [insignificant] coefficients (see Table 1 and Table 2 in Appendix). Furthermore, when put as a simple control outside of combinations, it also observes alike [insignificant] coefficients. Since the test does not yield any additional significant relationship featuring a dominant party, we are to infer that upon the circumstances laid out in the interaction terms, this factor actually adds to the likelihood of coup success. Once is replaced with , one extra term from Hypothesis 1 is corroborated at the 0.1 level. According to it, expenditure per soldier increases the outcome probability when grows. Yet as known from the previous section, this economic predictor acts in the opposite direction if coupled by . Such transformation may allude to the effect of a cumulative estimate present in the term, i.e. . Its upper values denote a greater threat «accrued» to the moment of a bid, implying that a government has been periodically subject to attempted ousters during the time in office. In other words, the army persists with subversion, and authorities get finally displaced. Recalling that it is a double interaction not involving the ethnic stacking variable, we can deduce that even personalized financial benefits is quite an unreliable measure if the threat exists and, moreover, occasionally reminds about itself. (Contrary to high , upon which enhanced funding becomes enough for lowering coup risk, in this sense surely demands manipulations with troops' composition.) Resources then turn into a means for ejection rather than appeasement, as suggested earlier. Finally, there is one double term which seems to contradict the effect of economic coup-proofing under high : it is the latter in combination with the change in military expenditure, evidenced at 10% in two models involving , , and either or . Though these indicators contribute to the term's significance, the more interesting question is why the sign is positive throughout all the models featuring such interaction. A tentative inference could be that enhancement of «untargeted» resources leads to their diffusion, which advances overall institutional capacity and, subsequently, capacity to intervene. However, such reasoning is difficult to defend in view of the antecedent results for and per soldier funding as well as the general idea of financial incentives.

Table 2. Regression Results

	Dependent variable:
	Successful coup
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Ethnic stacking	-2.334*	-2.416*	-1.253	-1.212	-1.241	-1.194	-2.367	-2.402
	(1.396)	(1.424)	(1.539)	(1.552)	(1.552)	(1.565)	(1.645)	(1.625)
Military regime ...

Подобные документы

• Политические технологии в социальных интернет-сетях на примере выборов в США 2012 года

  Теоретический анализ научных концепций и подходов к определению политических технологий. Интернет как фактор трансформации политической коммуникации. Социальные сети и медиа в президентской компании 2012 г. Краткая характеристика интернет-аудитории.
  
  дипломная работа [218,5 K], добавлен 28.04.2015
  

• Політичний режим Росії 2012 р.

  Передумови формування сучасного політичного режиму Російської Федерації. Погляди іноземних політологів на ситуацію в Росії. Президентство Володимира Путіна: режим "ручного управління" або "керованої демократії". Перебіг виборів Президента РФ 2012 року.
  
  реферат [30,0 K], добавлен 02.10.2013
  

• Стратегия и тактика избирательных кампаний на примере выборов во Франции 2012 года

  Понятие избирательной стратегии, ее функции и технологии. Выявление основных ресурсов избирательной кампании и тактических направлений. Анализ особенностей выборов во Франции 2012 года. Сравнение кандидатов в президенты Саркози и Франсуа Олланда.
  
  курсовая работа [49,6 K], добавлен 18.01.2015
  

• Democracy as a form of government

  The definition of democracy as an ideal model of social structure. Definition of common features of modern democracy as a constitutional order and political regime of the system. Characterization of direct, plebiscite and representative democracy species.
  
  презентация [1,8 M], добавлен 02.05.2014
  

• Трансформация политической системы в Республике Корея

  Образование Республики Корея, принятие первой Конституции. Формирование партийной системы. Попытки перехода от авторитаризма к демократии в период II-V Республик (1960-87 гг.). Становление и развитие политической системы современного корейского общества.
  
  курсовая работа [101,9 K], добавлен 11.06.2012
  

• Арабо-израильские войны 1960-х годов

  Арабо-израильские конфликты в канун "Шестидневной войны" 1967 года. Пакт Египта и Сирии о совместной обороне. Отношение СССР к войне. Начало операции "Голубь", установление израильского господства в воздухе. Советское присутствие в Египте, участие в боях.
  
  курсовая работа [47,2 K], добавлен 13.07.2009
  

• Public relations в государственном управлении, на примере пресс-службы Президента РФ

  Анализ структур, проблем и тенденций развития технологий Public Relations в системе государственной службы (на примере Управления пресс-службы и информации Президента). Ее основные задачи и функции. Предложения по улучшению функционирования пресс-службы.
  
  курсовая работа [316,8 K], добавлен 15.02.2016
  

• Проблемы институционального строительства

  Роль формальных политических институтов в недемократических режимах. Институциональная инженерия. Влияние манипуляций электоральными институтами на партийную систему. Сопоставление эффектов и последствий партийной реформы 2012 г. и корректировки 2014 г.
  
  курсовая работа [144,5 K], добавлен 15.12.2015
  

• Социально-политические конфликты в избирательной кампании 2011-2012 годов

  Анализ проблем формирования российской государственности на этапе перехода к постсоветской политической системе. Изучение степени влияния противоречий и конфликтов на процесс политического развития современной России. Борьба с терроризмом и экстремизмом.
  
  дипломная работа [68,5 K], добавлен 23.07.2015
  

• Франсуа Олланд

  Франсуа Олланде - победитель президентских выборов во Франции 2012 года от Социалистической партии. Карьера Франсуа Олланда до избрания президентом, его участие в президентских выборах. Родители, детство, семья и личная жизнь французского политика.
  
  реферат [96,6 K], добавлен 11.12.2012
  

• Політична реклама в Україні

  Політична реклама як основний чинник виборчої кампанії, її ціль і комунікативні завдання, особливості розвитку та застосування в Україні. Вплив ЗМІ на поведінку електорату. Маніпулятивні аспекти політичної реклами під час парламентських виборів 2012 р.
  
  курсовая работа [52,2 K], добавлен 23.01.2015
  

• PR на выборах в Верховную Раду Украины 2012 г. Удачи и неудачи предвыборных кампаний

  Сущность политической рекламы, ее функции и цели, средства достижения эффективности. Специфика политической рекламы в США и в Российской Федерации. Специфические черты PR на выборах украинской Верховной Рады, проблемы и успехи предвыборной кампании.
  
  курсовая работа [4,1 M], добавлен 30.03.2014
  

• Развитие внутренней и внешней политики Германии в течение 2005–2012 годы

  Экономическое развитие Федеративной Республики Германии. Кризис социального государства. Политическое развитие и внешняя политика ФРГ. Ангела Меркель – первая женщина на немецком Олимпе. Борьба за власть. Трансатлантическое направление сотрудничества.
  
  дипломная работа [126,1 K], добавлен 30.05.2015
  

• The political System of the United States

  Basis of government and law in the United States of America. The Bill of Rights. The American system of Government. Legislative branch, executive branch, judicial branch. Political Parties and Elections. Freedom of speech, of religion, and of the press.
  
  презентация [5,5 M], добавлен 21.11.2012
  

• Конвенциональные формы политического участия: особенности митингового движения в современной России

  Сущность, формы, разновидности политического участия. Организационно-правовые условия проведения митингов. Особенности отношения молодежи г. Тамбова к массовым акциям протеста 2011-2012 гг. Отношение к роли партии "Единая Россия" и внесистемной оппозиции.
  
  курсовая работа [926,3 K], добавлен 24.11.2013
  

• Особенности политического пиара в избирательных кампаниях в Украине

  Методологический аспект исследования особенностей политического пиара в избирательных кампаниях. История возникновения Public Relations. Сущность понятия "выборы". Украинский электорат и его этнонациональные особенности как объект избирательного PR.
  
  курсовая работа [59,1 K], добавлен 12.08.2010
  

• Islam in the eyes of the West

  The rivalry between Islam and Chistianity, between Al-Andalus and the Christian kingdoms, between the Christian and Ottoman empires triggered conflicts of interests and ideologies. The cultural explanation of political situations in the Muslim world.
  
  реферат [52,8 K], добавлен 25.06.2010
  

• The Demise of Democracy

  Thus democracy and modernism are closely intertwined, each providing a driving force. Darwinism, Freudianism, Leninism and Marxism combined to throw doubt on traditional Western mores, culture and standards of behavior. Rights Without Responsibility.
  
  статья [20,3 K], добавлен 25.11.2011
  

• Электронное государство

  Понятие и концепция электронного государства. Соответствие понятия "электронное государство" английскому понятию e-government. Формирование "электронного государства" на рубеже веков. Новый этап развития конституционного государства, содержание законов.
  
  доклад [25,1 K], добавлен 15.04.2009
  

• Presidents of Kazakhstan, USA and United Kingdom of Great Britain and Northern Ireland

  N. Nazarbayev is the head of state, Commander-in-chief and holder of the highest office within of Kazakhstan. B. Obama II is the head of state and head of government of the United States. Queen Elizabeth II as head of a monarchy of the United Kingdom.
  
  презентация [437,6 K], добавлен 16.02.2014
  

• главная
• рубрики
• по алфавиту
• вернуться в начало страницы
• вернуться к началу текста
• вернуться к подобным работам