External Interventions in Post-Cold War Africa – 1989-2010

External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa E ter al I ter e tio s i Post-Cold War Afri a, – 1 RICARDO REAL P. SOUSA OBSERVARE, Autonomous University of Lisbon The study of external interventions in conflict management is critical and has implications for international relations and conflict theory. Quantitative studies of the relationship between external interventions and civil war have been prone to some conceptual limitations (understudied lowerintensity periods) and data limitations (unavailability of event battle death data). This article presents a new external interventions data set covering the period between 1989 and 2010 for Africa, building on the Regan et al. (2009) data set, which covers the period between 1945 and 1999. Novel features of this new data set are: the recoding of the overlap period; a broader range of categories of intervention, including UN and non-UN missions; and wider temporal scope, by extending the period of analysis to 2010, by lowering the civil war threshold to 25 battle deaths, and by starting the conflict period from the date of the first battle death in each civil war (based on UCDP GED version-1.5-2011). The advantages of the data set are illustrated with an analysis of the different effects interventions have on high- and lowintensity conflict periods. KEYWORDS Africa, civil war, data, external interventions Author’s Origi al Ma us ript To cite this article: Ricardo Real P. Sousa (2015) External Interventions in Post-Cold War Africa, 1989– 2010, International Interactions, 41:4, 621-647, DOI: 10.1080/03050629.2015.1028626 To link to this article: http://dx.doi.org/10.1080/03050629.2015.1028626 1 Research for this article was supported by a scholarship from the Portuguese Science and Technology Foundation of the Ministry of Science, Technology and Higher Education of Portugal in the context of the Operational Programme Human Potential co-funded by European Social Fund (SFRH/BD/44998/2008). Fieldwork was possible i the o te t of the proje t Mo itori g Co fli ts i the Hor of Afri a of the Ce ter of Afri a “tudies–ISCTE, Lisbon University Institute (PTDC/AFR/100460/2008), funded by the same Foundation and Programme. Support as also re ei ed fro the Eras us U i ersit Trust Fu d a d the proje t Applied Ad a ed “tudies i De elop e t fu ded the Calouste Gul e kia Fou datio a d the Ce tre for African, Asian and Latin American Studies of the Lisbon School of Economics and Management at the University of Lisbon. I would like to thank Andrea Vilán, João Correlo, Peter van Bergeijk, and interviewees from six African embassies in Addis Ababa, Ethiopia, for providing comments on the development of the data set. In particular I gratefully acknowledge Patrick Regan for his support. Any remaining errors are my own. Data, replication files and other supplemental material is available at https://sites.google.com/site/ricardosousa2000/research. 1 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Civil wars have been a recurrent type of conflict in the past few decades (Gleditsch, Wallensteen, Eriksson, Sollenberg, and Strand 2002). The great majority of conflicts have involved external interventions by bilateral, multilateral, a d o state a tors. I Rega , Fra k, a d A di ’s data set, 69% of conflicts in Africa between 1989 and 1999 involved some form of external intervention. It is generally accepted that external interventions have an important effect on conflicts. What is debated is how different types and targets of i ter e tio s affe t a o fli t’s duratio , i te sit , a d out o e. Illustrative of this debate are findings that, on the one hand, neutral, military, and economic interventions lengthen conflicts (Elbadawi and Sambanis 2000; Regan 2002), while, on the other hand, this effect has been attributed to the cases in which the interveners pursued their own agendas (Cunningham 2010). Other results suggest that if the support is military and biased in favor of the challenging group, it can shorten the conflict (Collier, Hoeggler, and Söderbom 2004). This data set allows a reanalysis of these mechanisms. The article presents a new data set on external interventions in civil wars in Africa. It builds on the Regan et al. (2009) data set but adds to it significantly in terms of compatibility with other data sets, time coverage, and event coverage. The advantages of this data set are to operationalize conflicts based on the Uppsala Conflict Data Programme—Georeferenced Event Dataset (UCDP GED), considering the periods when a conflict had high, low, and no intensity. It lowers the conflict threshold from 200 (Regan et al. 2009) to 25 battle deaths and further expands the conflict period by starting the analysis from the first battle death and continuing it up to the last event with battle deaths. By using the UCDP GED data set as a reference, it makes the analysis more comparable with other research and allows researchers to directly link to other data sets. Furthermore, while most other data sets focus either on military, economic, diplomatic, or UN and nonUN missions, this data set builds on the si gular ad a tage of Rega et al.’s data set in covering the main t pes of i ter e tio s. To Rega et al.’s a al sis of ilitar , e o o i , a d diplomatic interventions and their subtypes, this data set adds the subtypes of military and diplomatic sanctions and, more significantly, a new coding of UN and non-UN missions, with details of the mission mandate. This allows for an analysis of a specific type of intervention controlling for a broader range of other types of interventions. Besides increasing the event type coverage of Regan et al. (2009), this data set extends the original time period (1944–1999) to 2010, making it appropriate for analysis of the post-Cold War period (1989– 2010). Furthermore, within this period, it increases the number of interventions coded, in relation to both Regan et al. and the Regan and Meachum (2014) interventions data sets, which focus on lowintensity conflict. This data set covering all African conflicts after the end of the Cold War can be used to contribute to newer and more nuanced theories of the effects of interventions on the intensity and duration of a conflict or of the different effects of interventions in periods of low- and high-intensity conflict. For instance, it can allow us to understand how different combinations of bilateral and multilateral 2 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa interventions affect conflict. It may be that the effects of multilateral interventions are influenced by bilateral initiatives and vice versa. By using a conflict-month unit of observation, the data set is disaggregated so as to capture relevant escalatory and de-escalatory effects resulting from the initiation of each specific intervention, while it is suitable to do more aggregated analysis with the conflict-year unit of observation commonly used to access the effect of interventions on conflict duration. The effects of specific interventions may differ according to whether the conflict is at a stage of high or low intensity. For instance, a military intervention in a low-intensity conflict may have a stronger escalatory effect on conflict intensity than in a high-intensity conflict, where it has a smaller marginal effect on the military capacities of the parties. By considering the conflict period as a whole (from the first to the last event with battle deaths), an analysis can be made of the effects of interventions in low- or no-intensity periods, even when these periods occur between periods of high-intensity conflict. In the next section, I identify the main characteristics of the data set in a comparison with a set of alternative data sets. I then present in more detail the methodology and descriptive information of the new data set. The article concludes with the application of the data set to the investigation of the different effects of interventions in low- and high-intensity conflict. EXISTING DATA SOURCES Existing research looking into external interventions in civil wars has focused mainly on separate analyses of military, diplomatic, or peacekeeping interventions. There are only two data sets that analyze them all together. (See Table 1 for a comparison of data sets.) The dimensions most commonly analyzed are foreign military interventions that overtly involve troops on the ground, as these are traditionally considered to have the direst of legal and conflict consequences. Such is the case for the Military Intervention Data (OMID) by Tillema (1989) and the International Military Intervention Dataset (IMI) by Pearson and Baumann, updated by Kisangani and Pickering (2009). The UCDP External Support data set by Högbladh, Pettersson, and Themnér (2011) codes military support provided in 10 separate categories: troops, access to territory, access to military/intelligence infrastructure, weapons, material/logistics, training/expertise, funding/economic support, intelligence material, other forms of support, and unknown types of support. Some data sets focus on civil war but also provide information regarding overt external troop involvement in the conflict dyad. These include the Correlates of War (COW) (Sarkees and Wayman 2010); the UCDP Armed Conflict (Themnér and Wallensteen 2012), and the Harbom and Wallensteen (2005) data set (these data sets are not compared in Table 1). <Table 1> 3 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Separate initiatives have coded diplomatic intervention (more specifically, the subtype of mediation). Such is the case for the International Conflict Management (ICM) data set (Bercovitch 1999), the Civil War Mediation (CWM) data set (DeRouen, Bercovitch, and Pospieszna 2011), and the UCDP Managing Intrastate Low-Intensity Conflict (MILC) data set (Melander, Möller, and Öberg 2009).2 Economic interventions in the context of civil war, intended to affect the balance of capabilities of the parties as defined for this data set, have not been objects of significant research. These interventions are different from the humanitarian and development aid types of assistance, which have been researched in depth in other studies. The dynamics of peacekeeping interventions is a field of research in itself; data sets focus on the United Nations (UN) missions (for example, Doyle and Sambanis 2000) and sometimes include also non-UN missions (Fortna 2004; Mullenbach 2013; Sambanis and Schulhofer-Wohl 2007). This data set provides a new coding of UN and non-UN missions with details on the mandates of these missions. The data set with the broadest coverage of types of interventions available at this moment is Regan et al.’s Third Part I ter e tio Data, a global data set that covers the period from 1944 to 1999. For conflicts where at least 200 battle deaths were recorded in a year, it codes military, economic, and diplomatic intervention types and typifies a series of subtypes of intervention. Regan et al. (2009) is a conflict-month data set that merges Regan (2002), which deals with military and economic interventions, with a diplomatic interventions data set developed by Regan and Aydin (2006). The data set in Regan (2002) contains 150 conflicts, of which 100 involved interventions, corresponding to a total of 13,048 conflict-months, of which 1,038 involved i ter e tio s. Rega a d A di ’s (2006) data set has 13,243 conflict-months, involving 68 conflicts and 436 diplomatic interventions. These data sets include interventions that are partisan or neutral and provide detailed identification on the state and non-state interveners. The scope of types of intervention and the disaggregation to a conflict-month unit of observation in the Regan et al. (2009) data set are the unique characteristics that were the reason for making it the initial reference for the current data set. Nevertheless, its limitations in terms of not covering more recent periods (the data set ends in 1999) and the absence of lower-intensity conflicts and conflicts with less than 200 battle deaths per year have been overcome in this new data set. More recently, Regan and Meachum (2014), looking at periods of political instability, developed a data set whose structure is similar to that of Regan et al. (2009). These are periods in which a series of variables are used to identify countries as being at risk of civil war, in cases where the conflict did not reach the 1,000 battle deaths that is commonly used as the threshold for defining a case of civil war. This is similar to what is here referred to as low-i te sit o fli t. Rega a d Mea hu ’s 4 data set covers the world in the period between 1957 and 2007, with a total of 449 interventions coded, but for the overlap period and region with this data set (Africa between 1989 and 2007) they have coded fewer interventions than in this data set. For the overlap period and region, Regan and Meachum (2014) have 2 The International Crisis Behaviour (ICB) data set (Brecher and Wilkenfeld 1997, 2000) does not focus on civil wars, including them only when linked to international crises. 4 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa 7,881 conflict-months with 130 interventions, while this data set has 4,347 conflict-months with a total of 406 interventions. “pe ifi all , the o el features of this data set ersus Rega et al.’s are: the recoding of the period from 1989 to 1999 in order to ensure consistency of coding, resulting in an improvement in the number of interventions coded3; the coding of a new period, from 1999 to 2010; the inclusion of subcategories of military and diplomatic sanctions, and UN and non-UN missions4; the lowering of the death threshold for civil war from the 200 battle deaths of Regan et al. to the 25 of the UCDP/Peace Research Institute Oslo (PRIO) Armed Conflict data set; the further extension of the scope of lowerintensity conflicts by starting conflict periods from the first event with a battle death; and the inclusion of events that occurred during inactive years, based on the UCDP GED, version 1.5 (Melander and Sundberg 2011).5 Other main characteristics of the Regan et al. (2009) data set have been retained. A first characteristic is the conflict-month unit of observation. A second characteristic is the identification of the target of an intervention according to whether it is biased or neutral. A biased intervention occurs when the target of the intervention supports either the government or the opposition. A neutral intervention can be considered one if an intervention does not affect or is not intended to affect the balance of capabilities between the parties; these are normally undertaken to promote peace. A third characteristic is the recording of an entry in the data set for each intervention. As the information on the target of an intervention is a key dimension in the analysis of the theoretical framework used in several studies (Collier et al. 2004; Elbadawi and Sambanis 2000; Regan et al. 2009), this data set is structured to provide details on it. However, coding the target for each intervention has the drawback that the data structure is not a pure time series, as n interventions in one month will produce n entries for that month. This data structure has implications for the statistical analysis, but an aggregation of interventions in the same month can be easily performed, although it means losing information in a key set of variables of intervention. Overall this data set is more appropriate for use when the researchers are interested in controlling for the effect of one type of intervention versus other interventions, when they would like to analyze postCold War dynamics in Africa, when they need a disaggregation of the conflict-month unit of observation, 3 Improvements in data availability and search capabilities since 2000 (when the first data set on military and economic interventions was developed by Regan et al. [2009]) means that more information on events has been made available in an accessible format. For Africa and the overlapping period of the two data sets (1989–1999), Regan et al. coded 242 interventions, while this data set codes 336 interventions. See attachment with the intercoder consistency analysis for more details. 4 Högbladh at al. (2011) maintain that partisan and neutral interventions should be analyzed separately, and they ide tif differe t data sets for that purpose. The o sider this to e a li itatio of Rega et al.’s data set. Such a differentiation is made in other studies; for instance, in the specific data sets on peacekeeping, mediation, a d ilitar i ter e tio s like those prese ted i Ta le . This arti le follo s Rega et al.’s stru ture, as it allo s, within a single data set, for analysis of each type of intervention while controlling for other types. 5 Inactive years are periods during which the threshold of 25 battle deaths in a year is not reached, with or without deadly events occurring throughout the year. Although UCDP GED includes the geographical positioning of conflict events, such information is not used because, at this stage, it is not possible to link interventions to subnational units across the spectrum of conflicts. 5 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa when they require detailed characterization of interventions, or when they need to link with other UCDP-compatible data sets. THE DATA SET The data set focuses on Africa, the region with both the largest number of civil wars and the largest number of external interventions—for instance, in the form of peacekeeping operations. Although external interventions data sets normally have a global coverage, there are other data initiatives that also focus on Africa. This is the case for the UCDP GED, which focuses solely on Africa, and the Armed Conflict Location and Event Data Project (ACLED), which covers mainly African conflicts. The data set is constructed with a conflict-month unit of observation, focusing on civil wars using the UCDP/PRIO defi itio of a o tested incompatibility that concerns government or territory where the use of armed force between two parties, of which at least one is the government of a state, results in at least 25 battle-related deaths Gledits h et al. . I Afri a between 1989 and 2010, there were 42 conflicts in 30 countries that met these riteria. The date the o fli t starts is the first i ide e of the use of armed force by an organized actor against another organized actor ... resulting in at least one direct death in either the best, low, or high estimate categories ... for a spe ifi te poral duratio (Sundberg, Lindgren, and Padskocimaite 2010). These two definitions correspond to 5,582 conflictmonths, 1,845 of which involved deadly conflict. Because each intervention is coded independently, some months with more than one intervention have more than one entry in the data set, resulting in 5,788 data set entries.6 External interventions have been added to this conflicts-months structure. The definition of external interventions is adapted from Regan et al. (2009) and Rosenau (1968). I define external interventions as convention-breaking political, economic, or military (including UN and non-UN missions) actions in a ou tr targeti g the ou tr ’s authorit stru tures i support of the government, in opposition, or neutral) in order to influence the balance of power between the parties or the conflict process. The 6 The difficulties in determining civil war deaths have been amply documented (Lacina and Gleditsch 2005) and warrant caution on the part of researchers, especially when dealing with a criterion based on a low threshold of deaths. In this case, the threshold of one battle death is followed and justified on theoretical bases and based on the validity and reliability of the UCDP data set and in particular that of the specific project of the UCDP GED extending the UCDP Armed Conflict information. Theoretically, the data set focuses on the intensity of a conflict and less on its onset. One argument for using a 25 battle deaths threshold to identify a conflict in the UCDP Armed Conflict data set is that this figure offers a threshold high enough to ensure that a conflict exists, taking into account possible errors when recording casualties. If this threshold were lower, for instance five or 10 battle deaths in a year, one could incur an error in identifying conflicts based on a number close to the possible reporting error. This data set respects this cautionary criterion. Only UCDP conflicts that reached 25 battle deaths are coded, but then for such conflicts the data set considers useful the data on the conflict events in the years leading up to or with less than 25 battle deaths. This means that the initiation of conflict can be coded on the date the first battle death occurs, broadening the scope of analysis, since this date may be different from the date on which the conflict reaches 25 battle deaths. Following the same logic, inactive years are also coded. These are years with less than 25 battle deaths after the conflict has reached 25 battle deaths. See Hoglund and Oberg (2011), who document the UCDP experience and procedures. 6 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa intervention is made by a third party foreign to the conflict country, and this third party can be a state or a non-state actor.7 Convention breaking refers to a significant and temporary change in the normal course of relations between the countries. A main qualification for the convention-breaking and exceptionality criteria are the characteristics of the intervention and the fact that interventions occur during a conflict. In coding the external interventions, we observed an overall rule of identifying when they affect the balance of capabilities of the parties. This means that external interventions are coded with one entry in the month when they start for each unique combination of intervention type and target, although with the possibility of having more than one intervener. The target of an intervention can be in support of the government, the opposition, or neutral.8 The original definition of Regan et al. (2009) is adapted in two ways. One is the explicit inclusion of both UN and non-UN missions, as they meet similar criteria for military or diplomatic interventions. It can easily be seen that an enforcement mission is similar to a bilateral or multilateral military intervention and that a political mission can have characteristics of an ongoing mediation or forum. Nevertheless, in much of the literature there is an understanding that peacekeeping missions (which are the main types of these new missions being coded) occur during peacetime periods and thus fall outside of an analysis of interventions in civil war. This view is not supported by the facts, as within the conflict period there are 58 missions, 29 of which do not have an enforcement or political mandate. This high number can be attributed partly to the broadening of the conflict period. It also reflects the fact that even if peacekeeping is effective in sustaining peace—defined as not having more than 999 battle deaths (Doyle and Sambanis 2006; Fortna 2004)—it may be less effective in ending the low intensity conflict that this definition of conflict is intended to capture. The second adaptation was already implicitly made by Regan et al. and has to do with broadening the concept of interventions to also include political and neutral interventions. In total there are 576 external interventions, including military, economic, and diplomatic ones, and both UN and non-UN missions. This corresponds to an average of 13.7 interventions initiated per conflict. The collection of data on external interventions is based on the news search engine FACTIVA, which provides broad coverage of the top news sources on Africa.9 All of the entries of Regan et al. (2009) for 7 A list of state and non-state actors is provided in the code book. Non-state actors are multilateral organizations, specific multilateral groups, or international nongovernmental organizations. Support by other non-state actors such as refugees, diaspora groups, and foreign rebels are not considered. 8 If there is a change of government in a month when an intervention has started, two interventions are coded for the month in order to reflect the change in target (these cases are rare). The opposition groups are not differentiated in the coding. Also, new entries are coded when the same intervener uses other types of intervention in one month and when one type of intervention is recurrent in more than one month. 9 A FACTIVA search was conducted based on the following parameters: (1) articles with at least the word intervention; (2) those with at least one of the following keywords of the type and subtypes of interventions: troops, naval forces, equipment, aid intelligence, advisors, air support, military sanctions, grants, loans, nonmilitary equipment, expertise, credits, relief from past obligations, economic sanctions, mediation, forum, arbitration, 7 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa the period from 1989 to 1999 were double-checked.10 Whenever doubts emerged, academic case studies were used. For five entries, due to the unavailability of data (news or academic case studies), the UCDP External Support Dataset was consulted, and it is the single reference for these entries. Because the cases of UN and non-UN missions are widely documented in official sources and academic publications, the coding information is based mainly on the Web sites and official documents of the United Nations (UN), the Organization of African Unity (OAU), the African Union (AU), and the European Union (EU), alongside Heldt and Wallensteen (2007), Mays (2011), and the SIPRI Multilateral Peace Operations Database (SIPRI MPOD). The conflict-interventions entries coded were cross-checked with the Dynamic Analysis of Dispute Management (DADM) database of intrastate dispute information (http://uca.edu/politicalscience/dadm-project/dadm-data-sets/). The data are further validated by six military attachés from six African embassies in Addis Ababa, Ethiopia, in December 2012. The interviewees were those who answered a request sent out to all embassies representative of countries in this data set. Coincidentally, there was at least one interviewee from each African sub-region, and all confirmed the conflict-interventions narratives with very few changes.11 With the data set of Regan et al. (2009), particular attention was paid to ensuring inter-coder consistency, which is reflected in the high correlation and similar patterns and regression results for the overlapping conflict periods (see footnote 9 and the attachment on inter-coder consistency for details). The coded interventions have 1,639 references, with an average of 2.8 references per intervention. Fifty-eight percent of all the references come from news reports: Reuters News, BBC Monitoring, Agence France-Associated Presse, All Africa, Xinhua News Agency, Associated Press Newswires, the Financial Times; Dow Jones International News, BBC Timeline, Inter-press Services, The New York Times, The Guardian, The Times, The Independent–London, and several others with fewer than 10 references each. About 29% of the references are from academic data sets, namely: the UCDP Peace Agreement, Regan et al. (2009), UCDP External Support, the Dynamic Analysis of Dispute Management (DADM), and the SIPRI Multilateral Peace Operations Database. About 9% of references are from official sources, including the UN, the AU, the OAU, the EU, and NATO. The remaining 4% of the references are from academic case studies.12 There is a predominance of Anglophone sources.13 Table 2 resumes the main data structure. recall, offer, request; (3) between 1/01/1989 and 12/31/2010; (4) within subjects: economic news, international political economic organizations, political/general news, and selection of top stories/trends/analysis; (5) for the specific country (in region). 10 The final data set contains 196 interventions originally from Regan et al. (2009), representing 34% of the total of 576 interventions in this data set, with changes made to the characterization of interventions in 11 countries. Details are identified in the codebook and attachment on inter-coder consistency. 11 Also, the internal consistency of the data set (calculated variables) was rechecked by a research assistant. For a detailed account of the procedure, see the code book. 12 The most numerous references by type are: Reuters (289), BBC (135), Agence France Press (130), and All Africa (63). These account for 65% of news references. DADM is the main academic source with 43% of academic references (222), followed by Regan et al. (2009), with 39% of references (200). The UN is the main source of official documentation, with 82% of official references (139) (values in parentheses are the absolute number of references). 8 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa <Table 2> The distributions of the conflicts per sub-region are presented in Figure 1. The sub-regions with the most conflicts are Eastern and Western Africa.14 <Figure 1> The types and targets of interventions are identified in Figure 2. Diplomatic and military interventions are the most recurrent types of intervention, and most interventions are either neutral or in support of the government. <Figure 2> Interventions are of the following types and subtypes (number of interventions in parentheses):15 1. Military type (161), with a specification as to whether it involved troops (62), naval forces (two), equipment or aid (48), intelligence advisors (22), air support (10), or military sanctions (17); 2. Economic type (52), with a specification as to whether it involved grants (23), loans (three), nonmilitary equipment or expertize (zero), credits (one), relief from past obligations (10), or economic sanctions (15). 3. Diplomatic type (305), with a specification as to whether it involved a case of mediation (248), international forum (24), arbitration (zero), recall of ambassadors (one), offers to mediate by third parties that were rejected (18), requests for diplomatic intervention by one of the warring parties that were rejected (eight), or political sanctions (6). Figure 3 presents a timeline of the number of interventions by type of intervention. 13 The event search was not limited to a predefined set of news sources, and all results based on the search criteria are analyzed. This procedure did not prevent a predominance of Anglophone sources even if some sources are based on African news reports, such as BBC Monitoring or All Africa. See Chojnacki, Ickler, Spies, and Wiesel (2012) for a discussion of the effect of sources on the coding of civil war events. 14 These are the sub-regions defined by the United Nations Statistics Division. 15 The structure of types and subtypes of interventions builds on Regan el al. (2009). The subtypes with fewer interventions or, in one case, with no intervention have not been removed or regrouped. In this way more disaggregated information is provided on the subtypes of interventions identified in the coding process. 9 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa <Figure 3> Diplomatic interventions predominate until 1999, but since 2000 they have had a more balanced distribution alongside military interventions. Economic interventions are less common than diplomatic and military interventions throughout the period under analysis. The military interventions in 1998 were driven by the Democratic Republic of Congo conflict, and in 2006 they are driven by the Somalia conflict. Military interventions mainly involve troops, economic interventions grants, and diplomatic interventions mediation. The criteria used to determine whether an intervention is a peacekeeping operation or not are based on Heldt and Wallensteen (2007). A peacekeeping operation consists of the deployment of military troops and/or military observers and/or civilian police in target states; and a mandate established for the purpose of separating the conflict parties, monitoring ceasefires, maintaining buffer zones, and taking responsibility for the security situation between formerly, potentially, or presently warring parties and being neutral toward the conflict parties but not necessarily impartial toward their behavior. The mandate is formalized in multilateral agreements, peace agreements, and resolutions of the UN or regional organizations. This covers all the possible mandates considered, with two exceptions: prevention or political operations and enforcement missions. These are also included, as they equally meet the definition of external intervention. For this reason the reference in this category is to a broader category of UN and non-UN missions, instead of peacekeeping. Thirty-two UN missions and 26 non-UN missions are classified according to the mandate based on the definitions of Diehl (2008), Doyle and Sambanis (2006), and Heldt and Wallensteen (2007) (see Figure 4). <Figure 4> Missions classified as preventive or political operations are normally deployed prior to the outbreak of armed conflict. These can be fact-finding or mediation missions (cases of UN mediation or peacemaking without, however, a follow-up peacekeeping mission). Also included under this heading are cases of UN political and peacebuilding missions run through the Department of Political Affairs (as opposed to the Department of Peacekeeping Operations) and which are sometimes also deployed after periods of highintensity conflict. These are not normally considered peacekeeping missions and do not fit the definition 10 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa stated earlier. Nevertheless, they constitute interventions that are similar to diplomatic ones and more permanent than typical ad hoc short-term negotiating teams.1516 Observer missions are the ones with a specific mandate of monitoring, reporting, or observing. These missions usually do not have a large military component (their personnel numbering in the hundreds) and have very limited rules of engagement (they are often unarmed missions). Traditional peacekeeping operations are normally interposition missions and have one of the following mandates: to provide protection through interposition or the separation of conflict parties and maintenance of buffer zones, monitoring ceasefires, maintaining law and order, and disarming and demobilizing factions. These are normally lightly armed missions. If the only function is one of these, or even includes the provision of security and humanitarian assistance, it is considered a traditional peacekeeping operation. Multidimensional operations have at least two additional mandates beyond the protection a traditional peacekeeping operation provides. These dimensions include electoral assistance/monitoring, humanitarian assistance, and the training of local police or the security sector. These missions normally have a substantial civilian component for performing these duties and can include transitional administration. Enforcement missions are those in which the peace operation is authorized under Chapter VII of the Charter and/or that frequently involve large-scale combat operations against one or more of the parties. These missions are not based on consent, can be with or without transitional administration, and are deployed to create—rather than maintain— peace.17 The target of the intervention is identified in terms of whether it is in support of the government (139), in opposition (60), or neutral (377) (see Table 3), and up to six parties are coded for each intervention. Most interventions in support of the government or the opposition are of a military nature, while most neutral interventions are diplomatic. A series of characteristics of the interveners is identified. These characteristics include geographical origins—for instance, if one of the interveners is from the same subregion or from Africa, if it is the AU or UN, and which African sub-regions (through states) are involved in the intervention. The type of third party is demarcated according to whether it is a single state or group of states, a single International Government Organizations (IGO) or group of IGOs, other interveners or a 16 Because the conflict threshold was lowered to 25 battle deaths in a year and the conflict period starts from the first event with a battle death, this data set covers periods of very-low-intensity conflict not considered in civil war data sets. Thus, there are cases of preventive intervention that would not have been identified had the higher 1,000 battle death threshold been used as in other studies. 17 The first four mandates are based on Chapter VI— Pea eful “ettle e t of Disputes Co se t a d the fifth mandate on Chapter VII— Use of “a tio s or For e to “ettle Disputes E for e e t . The odi g is for the strongest mandate a mission had, increasing in strength from preventive to political, to observer, traditional peacekeeping, multidimensional operations, and enforcement missions. There is a difference between UN Peacekeeping and AU Peace support missions, but in this case the UN Peacekeeping definition is followed, and AU missions are analyzed accordingly. 11 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa combination of state(s), IGOs, and other interveners; and whether any intervener is a permanent member of the UN Security Council (P5) or an ex-colonizer. <Table 3> Although the top state interveners are France and the United States, with Italy in 10th place, all the other interveners are from Africa (Libya, Ethiopia, Tanzania, Kenya, South Africa, Zimbabwe, and Uganda) (see Table 4). This pattern is repeated in the breakdown by type of intervention, except for diplomatic interventions where African states are the top interveners, with Italy and France being the first non-African states, in sixth and seventh place respectively among the top 10. State missions are all by non-African states. <Table 4> For non-state interventions (see Table 5), the main intervener is the UN, representing about 33% of all interveners, and African-based organizations (about 51% of all interveners, with the AU, the Economic Community of West African States–ECOWAS, the Intergovernmental Authority on Development– IGAD, Southern African Development Community–SADC, the Arab League, and the Great Lakes Regional Peace Initiative in the top 10). The fifth-most relevant intervener is the European Commission (EC), which accounts for about 7% of all interventions. <Table 5> Military and economic interventions normally have the involvement of parties from outside the region. Most of these are interventions by the UN or EC, but diplomatic interventions have significant involvement from Africa-based organizations. The UN and EC comprise about three-quarters of all military interveners (38% and 36% respectively), with African-based interveners comprising one-tenth. The UN and EC are the main economic interveners (35% and 44% respectively). Diplomatic interventions are predominantly African (66% of all interveners), followed by the UN (22% of all interveners). In terms of UN and non-UN missions, most interventions have UN involvement (55% of all interveners), with a significant number of interventions having African organizations involvement (28% of all interveners), followed by the EC (11% of all interveners). 12 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Overall, among both state and non-state interveners, African actors comprise the majority of interveners in military and diplomatic initiatives, while economic interventions are mainly initiated with the involvement of interveners outside the region (France, United States, the UN, and EC), and missions mainly involve the UN. Figures 5 and 6 summarize the annual battle death data count together with the number of interventions of each type. Based on the UCDP GED (v1.5) data set, the number of battle deaths per event is aggregated to the conflict-month in the data set (based on the date of termination of an event) and further aggregated to the conflict-year for these figures. <Figure 5> <Figure 6> The lower number of battle deaths from the year 2000 onwards requires zooming in from that date onwards for a better visualization of the pattern. At this level of aggregation and for all conflicts, a correlation can be drawn between the frequency of the number of battle deaths and the number of diplomatic and military interventions, as well as between the number of diplomatic interventions and the number of military ones. Specific civil wars (that in the Democratic Republic of Congo in 1998 and Somalia in 2006) have led to a higher number of interventions, though not in the context of a similarly higher number of battle deaths. Economic interventions are not correlated with any other indicator, and their use decreased in the period under analysis. For UN and non-UN missions, there are a similar number of missions initiated and a similar temporal pattern, with a peak in 1993 followed by a decline up to the late 1990s when there was a resurgence, with a steady number of new missions established annually from 2002 onwards. Before continuing with an illustration of the novel uses of the data set, it is worth mentioning some of its limitations. The main shortcoming of the data set is that it is limited to Africa since the end of the Cold War. While this does limit its use, resource constraints prevented a broader temporal and geographical coverage. However, Africa is the region with the highest number and frequency of both civil wars and interventions, and the post-Cold War period is the most relevant for studies of the dynamics of interventions in a contemporary age. For these reasons, even with these limitations, the data set has the potential to contribute to theory development and testing. As mentioned before, the data set provides detailed information on external interventions, with a single entry for each intervention, even if two or more occurred in the same month. This allows for richer information on interventions, which are the focus of this data set. The downside is that in its current 13 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa structure, the data set is not a panel data set; nevertheless, if necessary the data set could be easily transformed into a panel data set by aggregating the information on interventions by month. Ideally, the interventions would be coded from the date of initiation to the date of termination. Unfortunately, information on the initiation of interventions, particularly of the non-mission type, is more abundant than that on when those interventions ended. The reason for this is that news reports are more consistent in providing information on when an event begins than when it ends. For this reason, the data set codes all interventions by the date of initiation and, whenever possible, identifies the date when the interventions ended. Also, information regarding the size of the interventions, being measured in terms of soldiers, equipment, days, or money, could not be gathered for all interventions. This means that interventions with different sizes have the same weight in the data set. These two limitations apply to all types of intervention except to the missions, for which such information is coded systematically. Finally, due to resource and information limitations, it was not possible to use the conflict dyad and differentiate which opposition group an intervention targeted. DIFFERENT EFFECTS OF INTERVENTIONS IN HIGH- AND LOW-INTENSITY CONFLICT One of the advantages of the data set is that it allows a differentiation of the effects that interventions have in periods of high- and low-conflict intensity and on conflict escalation and de-escalation. The latter effects have not been investigated in previous studies, which have focused instead on conflict duration. Previous research on high-intensity conflict (above 200 battle deaths) has found that military and economic interventions increase conflict duration (Elbadawi and Sambanis 2000; Regan 2002), while diplomatic interventions can facilitate the termination of civil wars (Regan et al. 2009). Here duration is considered a proxy for intensity, as per the definition a conflict persists if it has a minimum threshold of battle deaths in a certain time period. Research on low-intensity conflict has found that military interventions increase the likelihood of a conflict reaching the level of civil war (by reaching the threshold of 1,000 battle deaths in a year), while economic and diplomatic interventions decrease its likelihood (Regan and Meachum 2014). Considering the theoretical underpinnings of these studies and applying them to the effects of external interventions on conflict intensity, it is possible to identify the following expected mechanisms for highand low-intensity conflict. It is generally expected that external interventions increase the capacity of the parties in a conflict. Military and economic interventions, which are normally partisan interventions, increase the fighting capacity of the parties and are expected to lead to an increase in conflict intensity. This effect is expected to be more significant in low-intensity periods, as additional capacity in high-intensity periods will make a smaller marginal contribution to the o fli t parties’ e isti g ilitar apa it . Diplomatic interventions, which are normally neutral interventions, increase the capacity of the parties to reach and sustain a political solution and are expected to lead to decreases in conflict intensity. This effect is expected to be relatively more significant in high-intensity periods than in low-intensity ones, as 14 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa any success of diplomatic interventions will be felt in the de-escalation from higher levels of battle deaths. These two mechanisms are tested with the following conflict intensity model: Conflict intensity is the count of battle deaths in a conflict i in a month t. The external interventions Iit have been operationalized, with an increasing effect during a period of 2 years after they have started. The rationale is that there is an absorptive capacity of the parties to take advantage of the support provided by the interveners. External interventions have a value of 1 in the first year and of 2 in the second year (with the exception of the subtype of military air support, which has a value of 2 in the month when it occurs). A vector of w control variables Cit relevant to either the conflict or the interventions is identified, and these are a categorical variable of the conflict intensity levels in the previous year,18 the natural log of the populatio , the pre ious ear’s pur hasi g po er parit adjusted GDP per capita, a proxy for the level of democracy in terms of the mode of executive selection,19 whether or not there are natural resources in the form of oil or gas, and the natural log of overseas development assistance (ODA). The model also includes a conflict-specific time trend starting from the initiation of the conflict (ditmontht), country fixed effects (ai), and a disturbance term (eit). The model is used in four regressions, presented in Table 6. The first regression is for all conflict-months, using a zero inflated negative binomial (ZINB) model (column 1 and 2).20 The second regression, using a zero inflated poisson (ZIP)21 model, is for high-intensity conflict-months, which are the conflict-months that occur in years when there were 1,000 or more battle deaths (columns 3 and 4). The low-intensity conflict-months are those that occur in years where there are fewer than 1,000 battle deaths. These low-intensity conflict-months are analyzed with a ZINB separately, in terms of their occurring before (columns 7 and 8) or after (columns 5 and 6) the conflict reaches 1,000 battle deaths per year. The results of the base ZINB model using all data are in line with results by the other researchers identified previously (model number in parentheses). Military and economic interventions have a strong escalatory effect in both increasing the likelihood of higher battle deaths (1) and decreasing the chances 18 Zero for zero battle deaths, 1 for between one and 24 battle deaths, 2 for between 25 and 999 battle deaths, and 3 for above 999 battle deaths. 19 One—direct election; two—indirect election; and three—non-elective. 20 The ZINB is justified on the grounds that zero is the modal category of the outcome variable, with 3,808 entries out of a total of 5,787 entries (66%), and the variance of battle deaths is higher than its mean for all data, for each conflict, and for each country. The ZINB model deals with the excess zeros in the count of battle deaths in two parts. A negative binomial model predicts the count of the cases with no certain zero battle deaths (int_m), and a logit model deals with the excess zeros (inflate), where an outcome of 1 means zero battle deaths and zero means some positive number. This estimation reflects the situation where there are no battles in a month (the excess zeros in the logit model), and the situation where there are battles in a month, whether they result in battle deaths or not (the no certain zero battle deaths in the negative binomial model). 21 The ZINB estimation was indeterminate, and ZIP is the second-best estimation for the data. The ZIP models use a similar procedure to the ZINB but make use of a poisson count model instead of a negative binomial model to predict the number of cases with no certain zero battle deaths. 15 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa of there being a month with zero battle deaths (2). Diplomatic interventions have a de-escalatory effect, in decreasing the likelihood of higher battle deaths (1).22 The differentiation between high and low intensity provides a more nuanced account of the mechanisms. Military interventions have an escalatory effect (1, 2), but this effect is only realized in periods of lowintensity conflict that occur after the conflict has reached high intensity—1,000 battle deaths (5). Military interventions in high-intensity periods have no significant escalatory effect in months with a propensity for battle deaths (3). At the same time, in both high and low-intensity periods, military interventions decrease the likelihood of a conflict having zero battle deaths (4, 6). This is not necessarily an escalatory effect, as it could instead mean the maintenance of a certain level of intensity. Surprisingly, military interventions have no effect in low-intensity periods prior to the conflict reaching 1,000 battle deaths (7, 8). This could be explained by the idea that interveners do not commit themselves militarily to a conflict before it starts, but once it is ongoing, external parties assume the task of providing support to one or more of the parties in the conflict. Economic interventions have an escalatory effect for all conflict periods (1, 2) but no distinctly significant effect in high- or low-intensity periods (3 to 6), with the exception of economic interventions in lowintensity periods prior to a conflict reaching 1,000 battle deaths (8). In this case, economic interventions increase the likelihood of there being a month with zero battle deaths. This could suggest that economic interventions in the early stages of a conflict delay the escalation of conflict. Diplomatic interventions have surprising results. For all conflict periods, diplomatic interventions decrease the intensity of conflict (1), and this effect is stronger for low-intensity periods after they have reached 1,000 battle deaths (5) than for high-intensity conflicts (3). Surprisingly, in low-intensity periods prior to a conflict reaching 1,000 battle deaths, diplomatic interventions have no significant effect in deescalating the conflict (7), although they increase the likelihood of there being zero battles in a month (8). Also, diplomatic interventions in high-intensity periods contribute to a decreased likelihood of there being zero battle deaths (4). Overall, the expected stronger de-escalatory effect in periods of high intensity is not confirmed, even though diplomacy becomes more de-escalatory once a conflict has passed the 1,000 battle death threshold, particularly in low-intensity periods. <Table 6> These surprising results indicate that diplomacy does not always have a de-escalatory effect, perhaps because it is driven by the agendas of external supporters (Cunningham 2010). Nevertheless, it may also be the case that the reverse causality in the relationship between external interventions and conflict 22 UN and non-UN missions are not analyzed, although they have been included in the model in order to control for their effects. 16 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa intensity affects the results. We may wonder whether external interventions are a result of higher conflict intensity or higher conflict intensity is the result of external interventions. Although previous levels of conflict are controlled for in this model, there is a need for more elaborated analysis controlling for this endogeneity. CONCLUSION This article presents a new data set of external interventions that makes a contribution to the available data on external interventions in conflicts. It is the sole data set that covers the entire post-Cold War period in Africa and does so in terms of the most significant types of intervention: military, economic, diplomatic, and UN and non-UN missions. Also, it disaggregates interventions with information on intervention subtypes and mission mandates. Consequently, it allows us to analyze the effects of a specific type or subtype of intervention, while controlling for the effects of other interventions within a single data set and thus increasing the reliability of the analysis. Furthermore, the characterization of each intervention is extensive, covering intent, target, and interveners in terms of identification, location, and categorization, and with identification of the diplomatic outcome in the form of peace agreements. In terms of the time dimension, by using the conflict-month unit of observation, our use of disaggregation enables us to account for the variations relevant to a study of external interventions with both high- and low-intensity levels of conflict. An illustration is provided of how the novel characteristics of the data set can be used to test and develop newer and more nuanced theoretical explanations of the effects of external interventions on conflicts. 17 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa TABLES AND FIGURES Table 1 Comparison of intervention datasets Dataset (*1) External interventions in Africa, Sousa (2014) Regan et al. (2009) (*3) Temporal coverage 1989-2010 Types of interveners State, nonstate and multilateral State, nonstate and multilateral State Unit of observation Types of interventions Type of violence (*2) Civil war threshold Conflict-month (5582) Military, economic, diplomatic, UN and nonUN mission (576) Military, economic, diplomatic (1474) Civil war (UCDP) (42) 25 battle deaths in a year, from the first to the last event with at least one battle death 200 battle deaths in a year Conflictintervention (269 conflicts) Military (591) International armed conflict 1989-2005 State and multilateral Conflictintervention-event (1115) Military Political issue or dispute Primary yearly warring party receiver of support in a conflict dyad (3606) Conflict (309 conflicts) Military Civil war (UCDP) 25 battle deaths in a year Diplomatic (3377 conflict management events) International disputes (inter-state and civil internationalized disputes involving state and nonstate actors) Civil wars (UCDP) Conflict (no threshold) 1944-1999 Conflict-month (13486) Civil war (150) Military Intervention Data (OMID), Tillema (1989) International Military Intervention Dataset (IMI), Kisangani and Pickering (2009) UCDP external support dataset, Högbladh et al. (2011) 1945-1985 1975-2009 State and non-state groups International Conflict Management (ICM), Bercovitch (1999) 1945-2000 State, nonstate and multilateral Civil War Mediation (CWM), Karl DeRouen et al. (2011) UCDP Managing Intrastate LowIntensity Conflict (MILC), Melander et al. (2009) Fortna (2004) (*4) 1946-2004 State, nonstate and multilateral Civil wars (319 civil wars) Diplomatic (460 mediation events) 1993-2004 State, nonstate and multilateral Diplomatic and UN and non-UN peacekeeping (3018 events) Low level armed conflict where one side is the government (UCDP) Between 25 and 1000 battle deaths in a year 1944-1997 Multilateral Conflict dyad (127 conflict dyads with 83 having some form of third party intervention) Peace periods in or UN and non-UN missions Civil war 1000 battle deaths in a year 25 battle deaths in a year 18 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Dataset (*1) Sambanis and Schulhofer-Wohl (2007) (*5) Third-party Peacekeeping in Intrastate Disputes, Mullenbach (2013) Regan and Meachum (2014) Temporal coverage Types of interveners Unit of observation Types of interventions Type of violence (*2) Civil war threshold UN and non-UN missions (79) Civil war Intrastate disputes (280) 500 battle deaths in a year, or 1000 battle deaths within three years Dispute cases including precrisis, crisis, conflict, postconflict, and post-crisis. 1945-1999 Multilateral after civil wars (115 cases) Civil wars (145) 1946-2012 States and multilateral Peacekeeping in disputes Peacekeeping (202) 1957-2007 State, nonstate and multilateral Country-month (124764) Military, economic and diplomatic (449) Countries at risk of civil war (1000 battle deaths in a year) Notes: *1 - All datasets have a global coverage except the one presented in this paper. Where appropriate the number of entries of each dimension is identified in parentheses. *2 - UCDP/PRIO definition of a civil war is a “contested incompatibility that concerns government or territory where the use of armed force between two parties, of which at least one is the government of a state, results in at least 25 battle-related deaths” (Gleditsch et al., 2002). Doyle and Sambanis (2000, 2006) definition of civil war involves several criteria but, in essence, it is similar to the one of UCDP/PRIO in that it refers to an incompatibility involving the state and organized group(s), with the use of armed force surpassing a threshold of battle deaths within a certain time period. *3 - Regan et al. (2009) merges the Regan (2002) dataset on economic and military interventions with the Regan and Aydin (2006) dataset on diplomatic interventions. *4 - Because Fortna (2004) is based on Doyle and Sambanis (2000) the later is not identified in the table. * 5 - Sambanis and Schulhofer-Wohl (2007) is based on Doyle and Sambanis (2006). 19 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Table 2 Descriptive indicators of the dataset Conflicts in Africa 30 countries with 42 conflicts Period 5,582 months with 5,787 entries CW events Interventions 291.648 battle deaths occurring in 576 interventions 1,845 months Figure 1 Conflicts per sub-region Figure 2 Numbers of types and targets of interventions 20 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Figure 3 Military, economic and diplomatic interventions per year23 Figure 4 UN and Non-UN missions per year Table 3 Target of support by type of intervention Type of intervention Government Opposition Neutral Military 102 40 19 Totals 161 Economic 23 17 12 52 Diplomatic 8 2 295 305 UN Mission 1 0 31 32 Non-UN Mission 5 1 20 26 139 60 377 576 Total 23 Interventions are coded in the month of initiation, and in this figure are summed up in the corresponding year. 21 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Table 4 Top 10 state interveners per intervention type Code Country / Intervention Type All types Military Economic Diplomatic Missions 220 France 53 25 14 12 2 2 United States of America 34 14 9 9 2 620 Libya 25 10 1 14 0 530 Ethiopia 23 13 2 8 0 510 Tanzania 21 1 2 18 0 501 Kenya 19 2 2 15 0 560 South Africa 18 2 0 16 0 552 Zimbabwe 17 4 0 13 0 500 Uganda 16 7 2 7 0 325 Italy 15 1 1 13 0 483 Chad 15 6 0 9 0 540 Angola 12 9 0 3 0 625 Sudan 12 6 0 6 0 531 Eritrea 7 6 0 1 0 200 United Kingdom 6 5 1 0 0 517 Rwanda 8 4 2 2 0 651 Egypt 10 2 0 8 0 481 Gabon 8 0 0 8 0 Note: The table shows the top 10 state interveners for each type of intervention identified in bold. This means that in the first column of all types of interventions the first 10 states are the top interveners. In the second column the top 10 state military interveners include five that are also in the top 10 of all types of intervention (France, United States of America, Libya, Ethiopia, Uganda), plus an additional five countries (Angola, Chad, Eritrea, Sudan, United Kingdom). In economic interventions several countries are within the top ten with only one intervention; for these cases, only countries that are the top ten in other types of intervention are identified. The number of interventions for non-top 10 countries listed in the table is also reported. The same procedure is followed for economic, diplomatic and UN and non-UN missions. These criteria mean that some well-known interveners may not be listed because they are not top ten in any of the types, as in the case of Nigeria. Table 5 Top 10 non-state interveners per intervention type Code Non-state actor / Intervention Type Military Economic Diplomatic Missions United Nations (UN) All types 100 1.1 11 8 49 32 1.4 African Union (AU) 46 1 0 35 10 1.9 Economic Community of West African States Monitoring Group (ECOWAS/ECOMOG) Intergovernmental Authority on Development (IGAD) European Commission (EC) 43 1 1 36 5 26 1 0 25 0 22 8 7 1 6 Southern African Development Community (SADC) Arab League 11 0 0 10 1 10 0 0 10 0 10 0 0 10 0 3.7 Great Lakes Regional Peace Initiative on Burundi Catholic Church 9 0 0 9 0 4.1 Carter Center 8 0 0 8 0 1.5 1.3 1.6 2.4 2.7 22 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Code Non-state actor / Intervention Type All types Military Economic Diplomatic Missions 3.1 Inter-Congolese Dialogue 5 0 0 5 0 1.7 4 0 1 1 2 3.6 Economic Community of Central African States (ECCAS) Union of Soviet Socialist Republics (USSR) 3 2 0 1 0 1.12 World Food Programme (WFP) 2 0 2 0 0 3.5 Paris Club 2 0 2 0 0 1.14 World Bank (WB) 1 0 1 0 0 1.2 North Atlantic Treaty Organisation (NATO) 1 0 0 0 1 3.3 Joint Monitoring Mission/Joint Military Commission World Vision International 1 0 0 0 1 1 0 1 0 0 4.5 Note: The table shows the top 10 state interveners for each type of intervention identified in bold. In Table 4 it should be noted that for non-state interveners using military intervention several actors that are in the top ten conducted only one intervention. Also, here only the actors that were among the top ten of other types of interventions were coded. A noticeable absence in this table is the Commonwealth of Nations, formerly known as the British Commonwealth, which was only in the top ten of military interventions with one intervention, and therefore did not make it to the table, although it conducted three diplomatic interventions. Figure 5 Battle deaths in civil wars and types of interventions Note: Battle deaths on the left axis and all types of intervention on the right axis. Figure 6 Battle deaths in civil wars and types of intervention after 2000 Note: Battle deaths on the left axis and all types of intervention on the right axis. 23 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Table 6 Effects of interventions in conflict intensity in periods of high and low intensity ZINB all periods VARIABLES Military (1) int_m ZIP high intensity periods ZINB low intensity after high intensity periods (2) inflate (3) int_m (4) Inflate (5) int_m (6) Inflate ZINB low intensity before high intensity periods (7) int_m (8) Inflate 0.0214* -0.115*** -0.000347 -0.141*** 0.0643*** -0.0597** 0.0466 -0.0397 (0.0121) (0.0188) (0.0176) (0.0524) (0.0164) (0.0245) (0.116) (0.112) Economic 0.0506* -0.109** 0.0229 -0.0209 0.00856 -0.0598 0.185 0.245* (0.0270) (0.0445) (0.0599) (0.0983) (0.0340) (0.0787) (0.129) (0.140) Diplomatic -0.0405*** 0.00235 -0.0254* -0.0651** -0.0497*** -0.00256 0.0207 0.574*** (0.0110) (0.0131) (0.0153) (0.0304) (0.0156) (0.0195) (0.109) (0.134) -0.205** 0.0457 0.0935 0.350* -0.257** 0.198 0.0483 -0.534*** (0.0890) (0.0818) (0.169) (0.209) (0.101) (0.134) (0.175) (0.207) -0.152* 0.0895 0.0956 0.0179 0.0505 -0.271* -0.925*** -0.460 (0.0915) (0.104) (0.177) (0.273) (0.0997) (0.154) (0.288) (0.291) UN missions non-UN missions Previous year levels of conflict Natural log of population Previous year GDP pc Executive selection Existence of oil and gas Natural log of ODA Conflict time trend Country FE 0.243*** -0.788*** -0.0479 0.0136 0.456*** -0.771*** 0.0944 -0.300*** (0.0538) (0.0482) (0.107) (0.172) (0.0800) (0.0930) (0.1000) (0.0817) -7.037*** 0.669 -1.477 -14.76*** -1.112 -4.829* 4.399** 0.639 (0.562) (0.472) (3.272) (4.721) (1.321) (2.563) (1.800) (1.408) -0.000560** -0.000562** 5.89e-05 0.000428 -0.00103*** -0.00159** -0.000726* -0.00122*** (0.000219) (0.000236) (0.000526) (0.000938) (0.000301) (0.000778) (0.000381) (0.000384) -0.366*** 0.160*** -0.477*** -0.341 0.129* 0.201* -0.0330 0.104 (0.0568) (0.0614) (0.140) (0.209) (0.0785) (0.117) (0.152) (0.142) 1.215*** -0.843* -0.102 0.515 0.341 0.343 0.506 -2.882** (0.357) (0.469) (0.464) (0.781) (0.501) (0.687) (1.233) (1.179) 0.000748 0.265** 0.193** 0.209 -0.0669 -0.335** 0.582*** 0.357 (0.106) (0.0882) (0.210) (0.259) (0.139) (0.165) (0.239) (0.168) 0.00673*** -0.00219*** -0.00401 0.0254*** 0.00191 0.00986* -0.0122*** 0.00111 (0.00100) (0.000820) (0.00645) (0.00932) (0.00297) (0.00568) (0.00298) (0.00225) Yes Yes Yes Yes Yes Yes Yes lnalpha Observations 5,154 Yes 0.709*** 0.212*** 0.163 (0.0426) (0.0630) (0.102) 5,154 881 881 1,914 1,914 2,359 2,359 24 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 25 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa REFERENCES Bercovitch, Jacob. (1999) International Conflict Management 1945–1995: Official Codebook for the International Conflict Management Dataset. Christchurch, New Zealand: University of Canterbury. Brecher, Michael, and Jacob Wilkenfeld. (1997, 2000) A Study of Crisis. Second Edition. Ann Arbor, MI: University of Michigan Press. Chojnacki, Sven, Christian Ickler, Michael Spies, and John Wiesel. (2012) Event Data on Armed Conflict and Security: New Perspectives, Old Challenges, and Some Solutions. International Interactions 38(4):382–401. Collier, Paul, Ank Hoeggler, and Måns Söderbom. (2004) On the Duration of Civil War. Journal of Peace Research 41:253–273. Cunningham, David E. (2010) Blocking Resolution: How External States Can Prolong Civil Wars. Journal of Peace Research 47(2):115–127. DeRouen, Karl, Jacob Bercovitch, and Paulina Pospieszna. (2011) Introducing the Civil Wars Mediation (CWM) Dataset. Journal of Peace Research 48(5):663–672. Diehl, Paul F. (2008) Peace Operations. Cambridge: Polity. Doyle, Michael W., and Nicholas Sambanis (2000) International Peacebuilding: A Theoretical and Quantitative Analysis. American Political Science Review 94(4):779–802. Doyle, Michael W., and Nicholas Sambanis. (2006) Making War and Building Peace: United Nations Peace Operations. Princeton, NJ: Princeton University Press. Elbadawi, Ibrahim, and Nicholas Sambanis. (2000) External Interventions and the Duration of Civil Wars. Policy Research Working Paper 2433, The World Bank Development Research Group: 1–18. Fortna, Virginia P. (2004) Does Peacekeeping Keep Peace? International Intervention and the Duration of Peace after Civil War. International Studies Quarterly 48(2):269–292. Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and Håvard Strand. (2002) Armed Conflict 1946–2001: A New Dataset. Journal of Peace Research 39(5):615–637. Harbom, Lotta, and Peter Wallensteen. (2005) Armed Conflict and Its International Dimensions, 1946–2004. Journal of Peace Research 42(5):623–635. Heldt, Birger, and Peter Wallensteen. (2007) Peacekeeping Operations: Global Patterns of Intervention and Success, 1948–2004. Second Edition. Research report. Sandöverken, Sweden: Folke Bernadotte Academy. Höglund, Kristine, and Magnus Öberg, eds. (2011) Understanding Peace Research: Methods and Challenges. London and New York: Routledge. 26 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Högbladh, Stina, Therése Pettersson, and Lotta Themnér. (2011) External Support in Armed Conflict, 1975–2009: Presenting New Data. Paper presented at the 52nd Annual International Studies Association Convention, Montreal, Canada, March 16–19. Kisangani, Emizet F., and Jeffrey Pickering. (2009) The International Military Intervention Dataset: An Updated Resource for Conflict Scholars. Journal of Peace Research 46(4):589–599. Lacina, Bethany, and Nils Petter Gleditsch. (2005) Monitoring Trends in Global Combat: A New Dataset of Battle Deaths. European Journal of Population 21(2–3):145–166. Mays, Terry. (2011) Historical Dictionary of Multinational Peacekeeping. Third Edition. Lanham, MD: Scarecrow Press. Melander, Erik, Frida Möller, and Magnus Öberg. (2009) Managing Intrastate Low-Intensity Armed Conflict (MILC), 1993–2004: A New Dataset. International Interactions 35(1):58–85. Melander, Erik, and Ralph Sundberg. (2011) Climate Change, Environmental Stress, and Violent Conflict: Test Introducing the UCDP Georeferenced Event Dataset. Paper presented at the International Studies Association, Montreal, Canada, March 16–19. Mullenbach, Mark J. (2013) Third-Party Peacekeeping in Intrastate Disputes, 1946–2012. Midsouth Political Science Review 14:103–133. Regan, Patrick M. (2002) Third Party Interventions and the Duration of Intrastate Conflict. Journal of Conflict Resolution 46(2):55–73. Regan, Patrick M., and Aysegul Aydin. (2006) Diplomacy and Other Forms of Intervention in Civil War. Journal of Conflict Resolution 50(5):736–756. Regan, Patrick M., Richard W. Frank, and Aysegul Aydin. (2009) Diplomatic Interventions and Civil War: A New Dataset. Journal of Peace Research 46(1):135–146. Regan, Patrick M., and M. Scott Meachum. (2014) Data on Interventions during Periods of Political Instability. Journal of Peace Research 51(1):127–135. Rosenau, James. (1968) The Concept of Intervention. Journal of International Affairs 22:165–176. Sambanis, Nicholas, and Jonah Schulhofer-Wohl. (2007) UN and Non-UN Peace Operations. In Multilateralism and Security Institutions in the Era of Globalization, edited by D. Bourantonis, K. Ifantis, and P. Tsakonas. London: Routledge. Sarkees, Meredith R., and Frank Wayman. (2010). Resort to War: 1816–2007. Washington, DC: CQ Press. Sundberg, Ralph, Mikael Lindgren, and Ausra Padskocimaite. (2010) UCDP GED Codebook version 1.0–2011. Department of Peace and Conflict Research, Uppsala University, Uppsala, Sweden. Themnér, Lotta, and Peter Wallensteen. (2012) Armed Conflict, 1946–2011. Journal of Peace Research 49(4):565– 575. 27 External Interventions in Post-Cold War Africa, 1989–2010 – Ricardo Real P. Sousa Tillema, Herbert K. (1989) Foreign Overt Military Intervention in the Nuclear Age. Journal of Peace Research 26(2):179–195. 28