Rebel governance: military boon or military bust? (Isard Award Article)

Abstract

What is the relationship between rebel governance and rebel military strength? Most existing research assumes that rebel governance enhances the military strength of the rebel group. I test this assumption with an original dataset of rebel governance services. The quantitative evidence presents a more complicated picture that belies a straightforward link between the two: governance appears to have either no relationship with rebel strength and sometimes even a negative and statistically significant relationship with rebel military capacity. To explain this surprising result, I generate a set of empirically grounded mechanisms using case vignettes that incorporate primary and secondary data. As a whole, the paper calls for greater theorizing and testing of the consequences of rebel governance, as well as the strategic motivations for its implementation.

Keywords

Civil wars insurgencies rebel governance rebel strength

Introduction

What is the effect of rebel governance on rebel strength? In its heydey, the Islamic State in Iraq and Syria constructed an elaborate amalgam of governance institutions, keen to establish a transnational caliphate during war (Revkin and McCants, 2015; Revkin, 2019, in press). However, the Islamic State’s military success and its brutality quickly provoked domestic and international backlash, as foreign powers intervened and supported domestic challengers to its rule. The international campaign against the Islamic State forced the organization to devote additional resources to its military efforts (Revkin and McCants, 2015). Given the intensified focus on preserving its hard-won victories, combined with a fixed and finite pool of resources upon which to draw, it turned to increasing or introducing new taxes to maintain its high quality of governance. In so doing, it provoked the ire of civilians (Revkin and McCants, 2015). At the same time, if it had never implemented or increased taxes but instead abandoned or reduced its governance efforts, experts argued that it would also would have lost civilian support for failing to provide the services that had become one of the organization’s hallmarks (Revkin and McCants, 2015).

The Islamic State faced a challenging paradox: increase taxes to maintain governance, but provoke civilian antagonism, or leave taxes unchanged, reduce governance efforts, and increase civilian antagonism. Furthermore, the Islamic State is far from the only rebel group forced to make difficult decisions about the very governance institutions that were supposedly critical to rebels’ ultimate victory. For instance, Kasfir (2005: 288) writes that when the National Revolutionary Movement (NRM) in Uganda was under increasing military pressure, it decided to largely abandon the governance institutions it had created for civilians in order to ensure the group’s military survival. Likewise, as the Revolutionary Front for an Independent East Timor (FRETILIN) faced increasing pressure from the Indonesian government, the military officers requested that civilians surrender and that FRETILIN abandon its governance activities: military (but not political) officers believed that the group’s governance efforts and the civilians living under FRETILIN’s control had increasingly become a liability to the rebel group (CAVR, 2005: 2–3, 21–22). Similarly, Stearns (2012: 212) recounts that although Ernest Wamba dia Wamba, one-time commander for the Rally for Congolese Democracy in the Democratic Republic of Congo, wanted to govern and even established health and transport departments for the provision of related social services, the costs of governance in addition to the costs of the military campaign were simply too burdensome for the group to undertake both activities. Finally, the activities of the governance wing of the United Liberation Front of Assam (ULFA) actually stoked “social tension” and “clashes” among the very people it was trying to govern. The confrontations became so extreme that one indigenous community established a blockade to fend off ULFA attempts to provide governance to them (Mahanta, 2013: 82–83).

That governance might impede, imperil or erode the militarily strength and popular support of rebel groups is somewhat surprising, given that much of the research on rebel governance suggests that the primary motivation for governance efforts is to reap military benefits (see, e.g. Arjona, 2016; Grynkewich, 2008; Huang, 2016; Lidow, 2016; Mampilly, 2011; Stewart and Liou, 2017; Weinstein, 2006). Rebels’ provision of governance is thought to curry favor and legitimacy with locals, which in turn allows rebels to more easily access information (Kalyvas, 2006) or needed resources like medicine, food, clothing or weaponry from civilians. Under the best of circumstances, the popularity that rebel leaders enjoy can even ease recruitment, thereby swelling the ranks of rebels. The influx of resources and recruits as a result of rebel governance efforts means that rebels are stronger, better equipped and ultimately better positioned to contest their enemies.

The anecdotes above, however, suggest that the relationship between rebel governance and rebel group strength is perhaps not as close as previously supposed. Rather than governance contributing to the military capacity of rebel groups, especially when rebels were most in need of greater resources and personnel, their governance efforts did little to assuage pressing military needs, and in some cases even hampered military efforts. For instance, the Islamic State’s high-quality and expansive governance should have eased its extraction of resources from the civilian population, especially resources dedicated to civilians’ own protection, but instead, the organization faced greater civilian discontent at the prospect of higher taxes. At the same time, when the NRM and FRETILIN abandoned governance in places still under their control, these organizations not only had to shoulder the sunk costs of first putting the institutions in place, but also had to forgo potential future profits associated with exploiting these governance institutions’ ability to attract new resources and recruits.

The contradictions between theoretical assumptions and certain empirical examples represent a puzzle I investigate here. Using original data, I test the assumption that rebel governance and rebel military capacity are positively associated. What I find is that the correlation between these two measures is weak and rarely positive, even sometimes statistically significantly negative. I then use qualitative data to inductively generate a set of mechanisms that might explain the null and at times negative relationship between rebel governance and rebel military capacity. In so doing, I highlight additional motivations for why rebel groups introduce governance, as well as some of the potential consequences of rebel governance on conflict dynamics and rebel group behavior.

Critically, this paper’s contribution is not to contradict or dispel the notion that rebel governance improves rebels’ military capacity. Rather, the importance of this article is to explore a central assumption that undergirds a growing research program, to identify a set of empirically grounded mechanisms that could account for the (lack of a) relationship between rebel governance and military capacity, to begin to understand the consequences of rebel governance and in doing so, expand the potential motivating factors for rebel governance in the first place. In the next section, I review how many existing works have approached rebel governance. I then use quantitative data from Stewart (2018) to evaluate the relationship between rebel governance and rebel military capacity. Then, I turn to qualitative research to generate a set of mechanisms that could explain the relationship between rebel governance and rebel military capacity. I conclude with a discussion of future applications to rebel governance.

Existing research

Drawing from scholarship on state formation and taxation (Levi, 1989; Olson, 1993), one of the most common and fundamental assumptions about why rebels govern is that governance is beneficial because its benefits enable rebels to better contest the state militarily. The logic behind this assertion is that, when rebels provide governance, they engender popular support and create legitimacy for themselves. This popular support and legitimacy enables rebels to more easily extract resources and recruits from the civilian population. Because rebels can easily extract resources and recruits from the population, they are stronger and better able to contest the state.

For example, Weinstein (2006: 163) writes that “[c]ivilians are thought to be central players in insurgency: access to food, shelter, labor, and information depends on their compliance. For this reason, rebel groups often build governing structures that mobilize political support from noncombatants and enable the extraction of key resources.”Lidow (2016: 8–9) builds on the work of Weinstein (2006) and argues that “[r]ebel leaders generally have an incentive to create security [a form of governance] in their territory because it allows for reliable access to food, military intelligence, and new recruits. Security also facilitates large-scale resource extraction, humanitarian aid operations, and other activities.” Similarly, Kalyvas (2006: 123–124) notes that providing benefits and protecting civilians from violence is one mechanism for securing civilian collaboration. Once again, rebel governance confers a military benefit to the rebel group undertaking these activities: the provision of governance (frequently in the form of security) enhances the group’s military resources and its overall size.

In a related vein, Huang (2016: see table 3, p. 76) finds that rebel reliance on civilians for economic support determines how much they invest in governance, and suggests that this could be because “rebels engage in statebuilding precisely where they depend on civilians for material and political support … Rebel statebuilding is a form of control, and enables rebel groups to elicit voluntary or coerced collaboration from the people under their authority” (Huang, 2016: 74). Arjona (2016: 50) also argues that “[e]stablishing a social contract … facilitates monitoring [of civilians] and makes both [civilian] obedience and spontaneous support more probable, which in turn favor[s] territorial control as well as its byproducts” that include “material contributions, political support, and recruits.”Grynkewich (2008: 353) notes that “non-state social welfare organizations offer the population an alternative entity in which to place their loyalty,” and that “a group that gains the loyalty of the populace commands a steady stream of resources with which it can wage battle against the regime.”Stewart and Liou (2017) and Mampilly (2011: 54) suggest that to gain the resources and recruits needed to survive, rebels could forcibly and coercively extract them from the populace, or use governance such that civilians willingly relinquished them. Mampilly (2011: 54) specifically argues that “the ability to provide a modicum of stability can be a powerful lure to civilians seeking refuge. And the provision of other benefits, including but not limited to the establishment of schools and hospitals, can provide a powerful incentive for civilians to support insurgent rule, even if only passively. From the perspective of the insurgent organization, reaching out to the larger noncombatant population makes tactical sense.”

The works discussed above share a similar underlying logic: rebels provide governance to strengthen their organization in ways that make them better able to contest the state. The reason for this is that governance is popular and desirable to civilians. Because civilians respond positively to governance, rebels are better able to secure civilian compliance to their authority, access information from civilians and in some cases even extract recruits and resources from them. In sum, rebel governance improves the overall military capacity of rebel groups and strengthens their hand when it comes to contesting the state.¹

Although the assumption of the militarily salubrious consequences of rebel governance is common in the literature, new and emerging scholarship has suggested that rebel governance institutions are not equally desirable to civilians, and in turn, are not as militarily beneficial as once supposed. For example Terpstra and Frerks (2017) argue that some governance interventions eroded the legitimacy and popular support of the Liberation Tamil Tigers of Eelam. More problematically, Kalyvas (2018), Arjona (2015) and Opper (2018) all note that some rebel governance interventions are either unpopular or actually provoke civilian violent resistance to rebel rule. Sometimes, this resistance is so troublesome that it imperils rebels’ control of territory or forces them to surrender altogether (Opper, 2018).

These recent works are far less sanguine about the military benefits of rebel governance. To better adjudicate between pre-existing theoretical assumptions and these recent empirical examples, in the next section, I use quantitative data to evaluate the military consequences of rebel governance. The pervasive understanding of rebel governance in the scholarly work is that governance enables the insurgency to better contest the state by increasing its capacity. This means that the provision of governance should be associated with stronger and more militarily capable rebel groups. Yet, recent work on cases of rebel governance suggests that the correlation between these two factors is perhaps less strong than previously supposed.

A quantitative assessment of rebel governance and rebel military strength

In this section, I conduct a preliminary analysis of the relationship between rebel governance and rebel military capacity. This analysis is explicitly non-causal. Its sole purpose is to determine whether a correlation between rebel governance and rebel military capacity exists, and if it exists, whether the implementation of rebel governance is associated with stronger rebel groups.

For this correlational test, I use replication data from Stewart (2018) which introduces a new measure of rebel social service provision, but also relies heavily on data from Cunningham et al. (2009). These data contain a variety of variables at the state and rebel group levels, including rebel governance and military capacity. This dataset is a cross-section of approximately 300 rebel groups that existed from 1945 to 2003, and the unit of analysis is individual rebel groups (from Cunningham et al., 2009).

The primary dependent variable of interest is Rebel Group Strength, which is an operationalization of rebel military capacity. The variable Rebel Group Strength is from the Non-State Actor Dataset from Cunningham et al. (2009). The ordinal measure has five unique values ranging from “0” to “+4,” with a “0” signifying the rebel group is much weaker than the state, and “+4” signifying the rebel group is much stronger than the state.

Conceptually, the primary independent variable of interest is rebel governance. In this paper, I operationalize rebel governance as rebel provision of education and healthcare activities, as well as the inclusivity therein. Although there are many ways to measure rebel governance—from taxation (Revkin, in press) to social services (Heger and Jung, 2017; Huang, 2016; Mampilly, 2011; Stewart, 2018) to elected bodies and executive institutions (Huang, 2016) —I focus on education and healthcare specifically. Unlike certain political institutions or taxation, both education and healthcare are broadly desirable and beneficial to all persons. In so doing, I avoid measures of potentially unpopular forms of governance (like taxes) that could unnecessarily bias against expectations that rebel governance facilitates military benefits. Additionally, healthcare and education are broadly similar across time and space: a basic literacy course provided in 1952 in Vietnam is essentially the same as a basic literacy course offered in 2001 in South Sudan. Other governance institutions, such as political institutions, might vary considerably across contexts, making observations incomparable in a global sample (Mampilly and Stewart, in preparation).²

Stewart (2018) collects data for both education and healthcare provision by rebel groups, as well as information about which categories of persons had access to these governance services. For instance, data from Stewart (2018) distinguishes between rebel groups that preserved social services for their own use internally and those that offered governance to civilians beyond the group. For instance, the Lord’s Resistance Army (LRA) provided medical training, but did not offer medical services to civilians outside the insurgency (Human Rights Watch/Africa and Human Rights Watch Children’s Rights Project, 1997: 27). In contrast, the African Party for the Independence of Guinea and Cape Verde (PAIGC) created elaborate and extensive governance institutions ranging from education (Dhada, 1993: 97–115) and healthcare (Dhada, 1993: 61–72) to elections (Dhada, 1993: 122–125, 205–207). These services were not solely reserved as selective incentives for fighters, but rather offered broadly to the civilian population. The PAIGC’s education program has even been described as “literacy for all” (Dhada, 1993: 97).

Using the cross-sectional replication data in Stewart (2018), I create a new variable signifying the provision of Any Services using the variable pg_ord. The variable pg_ord is an ordinal measure ranging in increments of “+1” from “0” to “2” with a “0” signifying no governance, a “1” signifying some governance, and a “2” signifying the provision of governance to people outside the rebel group’s primary coalition. The new variable I create, Any Services, is a binary variable with a “0” signifying no governance, and “1” signifying the construction of any forms of rebel governance.

As an additional test, I include the variable Most Inclusive Services from Stewart (2018). This binary variable is coded as “1” if rebel groups provide governance to persons outside their target coalition (e.g. non-coethnics, non-coreligionists) in addition to members of their target coalition (e.g. co-ethnics), active supporters and members, and “0” signifying otherwise. Importantly, this measure explicitly captures the provision of governance to people unlikely to join the rebel group, in addition to those who actively support the rebel group. This form of governance is thus a more costly investment unless rebels are able to convert services into recruits or resources.

Finally, consistent with Stewart (2018), in some of the tests below, I stratify my sample to include only rebel groups that control territory. The reason for this is because there could be some underlying factor that systematically determines both rebel strength and rebel governance, and by stratifying on territorial control, I ensure that units in the sample are relatively similar. One example is state capacity. Rebels operating in strong states might be unable to develop a serious military force, while also facing challenges to the provision of governance (Arjona, 2016). Territorial control is an excellent measure to use to stratify the sample because although territorial control is related to rebel strength and rebel governance, it is not the sole determinant thereof. For instance, rebel groups could control territory because a state is absent or too weak to challenge rebels, rather than because rebels have fended off state incursions. Likewise, territorial control is not necessary for the provision of any governance (Jackson, 2018): for example, rebels can provide literacy and mathematics classes to combatants and their families in highly mobile camps frequently hidden away in topographically challenging locales (Anderson, 1978).

To operationalize territorial control, I rely on the Territorial Control variable from Cunningham et al. (2009). The territorial control variable takes a value of “1” if a rebel group holds territory, and “0” if the rebel group does not hold territory. The left panel of Figure 1 presents the frequency of observations with any governance by various levels of rebel strength conditional on territorial control, while the right panel of Figure 1 presents the frequency of observations with the most inclusive governance by various levels of rebel strength conditional on territorial control.

Figure 1.

Governance by levels of strength, conditional on territorial control.

With these key variables, I conduct a preliminary analysis that investigates differences in the average (mean) level of rebel group strength among rebel groups that provide any form of governance, and rebel groups that do not provide any form of governance. The middle column of Table 1 presents the average (mean) level of rebel group strength for all groups that did provide some form of governance in comparison to all groups that did not provide any governance. Though not statistically significant, the findings are consistent with theoretical expectations: among all rebel groups, the organizations that provide any form of governance are, on average, stronger than the organizations that do not provide any governance.

Table 1.

Comparing rebel strength among governance-providing rebel groups.

	Rebel strength	Conditional on territorial control
Any Services	0.87	1.08
No Services	0.72	1.36
Difference of means	0.15	−0.28

Note: The table above presents the results of a two of difference of means tests comparing the average (mean) level of rebel strength of rebel groups that provide any services and rebel groups that provide no services. The second column investigates this difference between all rebel groups, while the adjacent, third column investigates this difference among all rebel groups that control territory. In the bottom row, entitled “Difference of means,” positive values indicate that rebel groups that provide any services are stronger than rebel groups that do not, while negative values indicate that rebel groups that provide services are weaker than those that do not. Among all rebel groups, rebels that provide any form of governance are slightly stronger, but these differences are not statistically significant. Among all rebel groups that control territory, rebels that provide any services are no stronger than rebel groups that do not, and may in fact be weaker. However, again, this difference is not statistically significant. *p < 0.10; **p < 0.05; ***p < 0.01.

When I limit the sample to rebel groups that control territory, however, the results are more surprising and counterintuitive. Among groups that control territory, those that provide any governance tend to be weaker on average than rebel groups that do not provide any governance. Again, this difference in the average strength between rebel groups that provide any governance and those that do not (conditional on territorial control) is not statistically significant. In Table 2, I repeat the analysis, but instead of investigating the average (mean) strength of rebel groups that provide any services compared with rebel groups that do not provide any services, I examine the average strength of rebel groups that provide the most inclusive services compared with rebel groups that do not provide the most inclusive services. The middle column of Table 2 presents the difference in the mean level of rebel strength among all rebel groups, and the third column of Table 2 limits the analysis to only rebel groups that control territory.

Table 2.

Comparing rebel strength among most inclusive governance-providing rebel groups.

	Rebel strength	Conditional on territorial control
Most Inclusive	0.84	0.93
Not Inclusive	0.78	1.24
Difference of means	0.06	−0.31*

Note: The table above presents the results of two difference of means tests comparing the average (mean) level of rebel strength of rebel groups that provide the most inclusive governance and rebel groups that do not. The second column investigates this difference between all rebel groups, while the adjacent, third column investigates this difference among all rebel groups that control territory. In the bottom row, entitled “Difference of means,” positive values indicate that rebel groups that provide the most inclusive services are stronger than rebel groups that do not, while negative values indicate that rebel groups that provide the most inclusive services are weaker than those that do not. Among all rebel groups, rebels that provide the most inclusive governance are slightly stronger, but these differences are not statistically significant. Among all rebel groups that control territory, rebels that provide any the most inclusive governance are weaker than rebel groups that do not, and this difference is not statistically significant at the 0.10-level. *p < 0.10; **p < 0.05; ***p < 0.01.

The results are similar to those found in Table 1. Among all rebel groups, organizations that provide the most inclusive governance tend to be stronger, on average, than those that do not. Again, this difference is not statistically significant. When I limit my sample to only rebel groups that control territory, however, the direction flips: rebel groups that do not provide the most inclusive governance tend to be stronger than rebel groups that provide governance at its most inclusive level. Interestingly, this difference becomes statistically significant at the 10% level.

Although the difference-of-means test illuminates important patterns in the relationship between rebel strength and rebel governance, regression analysis allows me to account for additional confounding factors that might influence rebel strength, beyond rebel governance or territorial control. Besides these three variables, I include several other factors that might influence rebel group strength. First, I account for several insurgency-level factors that might influence the strength of a rebel group. One potential factor is the goals of the rebel group, and so I include a binary variable coded as “1” if the rebel group is independence-seeking and a “0” if the rebel group is center-seeking, meaning that the group aims to overthrow the central government. The source data for this variable are from Cunningham et al. (2009).³ Additionally, ideology might also influence whether rebels elect to provide governance in the first place, so I include the variable Communist, which takes the value of a “1” if the rebel group espouses a communist ideology and a “0” otherwise (Mampilly, 2011).⁴ Because the number of people in a rebel group ostensibly affects its strength and capacity to challenge the state, I also include the variable Rebel Group Size. The variable Rebel Group Size is the natural log of the estimated number of people in the rebel group, from Cunningham et al. (2009). In Models 1 and 2 of Tables 3 and 4, I include as a control the variable Territorial Control (discussed above). In Models 3 and 4, however, I limit my analysis to only rebel groups that control territory.

Table 3.

Predicting rebel strength with any rebel governance.

	(1)	(2)	(3)	(4)
	Rebel strength	Rebel strength	Conditional on territory	Conditionalon territory
Any Governance	−0.019	−0.066	−0.135	−0.264
	(0.113)	(0.118)	(0.178)	(0.169)
Secessionist	−0.077	−0.022	0.214	0.439**
	(0.140)	(0.145)	(0.202)	(0.180)
Communist	−0.208*	−0.298**	−0.250	−0.336
	(0.119)	(0.122)	(0.192)	(0.199)
Rebel Group Size	0.214***	0.260***	0.271**	0.375***
	(0.044)	(0.055)	(0.112)	(0.111)
Territorial Control	0.286**	0.194*
	(0.122)	(0.102)
Competition	−0.018	−0.023	−0.048	−0.012
	(0.026)	(0.024)	(0.046)	(0.037)
Infant Mortality	0.001	0.001	0.002	0.002
	(0.002)	(0.002)	(0.003)	(0.004)
GDPpc	−0.023	−0.018	−0.018	−0.073
	(0.081)	(0.083)	(0.128)	(0.111)
Population (logged)	−0.165***	−0.155***	−0.277***	−0.357***
	(0.043)	(0.046)	(0.099)	(0.087)
Rugged Terrain	−0.015	−0.011	−0.015	−0.053
	(0.056)	(0.055)	(0.101)	(0.090)
Non-Military Aid		−0.170		−0.138
		(0.157)		(0.246)
Constant	1.970**	1.595*	3.603**	4.506***
	(0.778)	(0.834)	(1.590)	(1.341)
Observations	134	107	53	47
$R^{2}$	0.492	0.514	0.457	0.567

Note: OLS regression coefficients reported and standard errors are in parentheses. Standard errors are clustered by country. Across all models, the dependent variable in Table 3 is Rebel Strength, a five-value indicator variable. Positive coefficients indicate that independent variables are associated with higher levels of rebel group strength, while negative coefficients indicate that independent variables are associated with lower levels of rebel group strength. Models 1 and 2 present the results for all rebel groups in the sample for which data are available. Models 3 and 4 present the results for all rebel groups that control territory and for which data are available. *p < 0.10; **p < 0.05; ***p < 0.01.

Table 4.

Predicting rebel strength with most inclusive rebel governance.

	(1)	(2)	(3)	(4)
	Rebel strength	Rebel strength	Conditional on territory	Conditional on territory
Most Inclusive Governance	−0.349*	−0.394**	−0.628**	−0.786***
	(0.197)	(0.192)	(0.255)	(0.227)
Secessionist	−0.073	−0.010	0.341	0.616***
	(0.135)	(0.144)	(0.253)	(0.206)
Communist	−0.221*	−0.314***	−0.272	−0.318*
	(0.116)	(0.110)	(0.181)	(0.175)
Rebel Group Size	0.237***	0.279***	0.334***	0.468***
	(0.046)	(0.057)	(0.109)	(0.099)
Territorial Control	0.317**	0.233**
	(0.126)	(0.104)
Competition	−0.012	−0.021	−0.019	0.018
	(0.025)	(0.023)	(0.047)	(0.038)
Infant Mortality	0.002	0.002	0.003	0.002
	(0.002)	(0.002)	(0.003)	(0.004)
GDPpc	0.015	0.019	0.048	−0.007
	(0.082)	(0.085)	(0.119)	(0.102)
Population (logged)	−0.158***	−0.140***	−0.245**	−0.295***
	(0.041)	(0.044)	(0.105)	(0.080)
Rugged Terrain	−0.004	0.005	−0.017	−0.080
	(0.054)	(0.055)	(0.098)	(0.077)
Non-Military Aid		−0.162		−0.121
		(0.147)		(0.233)
Constant	1.283	0.786	1.847	2.001
	(0.848)	(0.935)	(1.823)	(1.437)
Observations	133	106	52	46
$R^{2}$	0.507	0.535	0.514	0.654

I also include several state-level variables that might influence rebel group strength. The first variable, Competition, is a count of the maximum number of rebel groups operating at the same time as an insurgency (coded from Cunningham et al., 2009). The more rebel groups there are, the greater the competition for resources and recruits between insurgencies is, and the higher the likelihood that rebels will turn against each other rather than the state, thereby weakening rebel groups. Three additional variables capture state capacity: Infant Mortality Rate the year conflict began, from World Bank (2012); GDPpc the year conflict began, from Heston et al. (2012); and Population (logged) the year conflict began, from Heston et al. (2012). Because it is more challenging for a rebel group to gain strength when facing an already formidable foe, a stronger state (with either a higher GDPpc or lower Infant Mortality Rate) is likely to confront weaker rebel groups. Likewise, if a state’s terrain is rugged and unnavigable, both rebel groups and states might have difficulty consolidating power over space and people. To account for this possibility, I include the variable Rugged Terrain from Fearon and Laitin (2003). Finally, in some models, I include a variable for whether a rebel group received Non-Military Aid (either political or financial backing) from Cunningham et al. (2009) and Högbladh et al. (2011).⁵ I include this variable in just a few models, however, because of the considerable missingness in the measure.

All models are estimated using ordinary least squares (OLS) regression with standard errors clustered at the country level. Table 3 reports the results for all models that include the variable Any Services, while Table 4 reports the results including the variable Most Inclusive Services. As a robustness check, I estimate the same model using ordered logistic regression (see Online Appendix Tables A.1 and A.2). Results are consistent despite the alternative estimator.

Many of the results are unsurprising. Rebel Group Size and Territorial Control are both positively correlated with rebel group strength, and this relationship is almost always statistically significant. In contrast, almost all measures of state strength indicate that it impedes rebel group’s capacity, and for Population (logged) this relationship is statistically significant. Although not statistically significant, the relationships between Rebel Strength and GDPpc, Infant Morality, Competition and Rugged Terrain are consistent with expectations.

What is surprising is the relationship between Rebel Group Strength and the two measures of rebel governance. Under no specification are rebel strength and rebel governance positively and statistically significantly related to each other, contrary to existing assumptions about the military benefits of rebel governance. Instead, across almost all specifications, the coefficient for rebel governance is negative, and in some models, this negative relationship is statistically significant, meaning that rebel governance is associated with lower levels of rebel strength. While these results are surprising, they are not inconsistent with recent statistical findings from other scholars. For instance, Balcells and Kalyvas (2015) find that socialist rebel groups, many of which introduce governance institutions because of their relationship to Maoist strategies (Mampilly, 2011), are more likely to be defeated and to experience long-lasting civil wars.

In the next section, I rely on case vignettes to present a set of mechanisms that could potentially explain the null and sometimes counterintuitive findings between rebel governance and rebel military capacity.

Potential explanations

The quantitative analyses above suggest that rebel governance and rebel strength are not closely correlated, and are sometimes even negatively associated. Given that the quantitative findings challenge fundamental assumptions about the consequences of rebel governance, in this section, I use case anecdotes from a variety of spatially and temporally diverse rebel groups to elucidate a set of potential mechanisms that might explain the findings from the quantitative analyses. The purpose of these proposed explanations is to rectify why little statistical evidence exists to support the claim that rebel governance results in military benefits for the rebel group.⁶ These mechanisms are also not mutually exclusive. Instead, these explanations are empirically grounded reasons for why rebel governance does not always enhance rebels’ military strength, or at least why there appears to be a weak, if not negative, relationship between rebel governance and rebel military capacity.

I argue that there are four broad baskets of potential mechanisms that could explain the findings in the previous section. The first explanation is that governance hinders, if not actively detracts from the military strength of rebel groups. The second explanation is that rebel governance actually improves the capacity of rebel groups, but the military benefits of rebels’ governance are limited or conditional. The third set of explanations is that the causal link between rebel governance and rebel military capacity has been misspecified. Finally, the fourth potential explanation is that data on governance or military capacity has been conceptualized, operationalized and collected in incomplete or problematic ways that do not allow for a statistical exploration regarding how these two variables are related.

Governance reduces the military capacity of rebel groups

One explanation for the null finding is that more governance negatively affects rebel groups’ military capacity. In this case, governance burdens rebel groups, such that some rebel groups persevere militarily despite providing governance. Furthermore, if governance is unpopular and requires rebels to divert ever more personnel from military activities to oversee the implementation of governance, chains of command could dissipate, leading to military indiscipline (Green, 2018; Parkinson and Zaks, 2018).

Although rebel governance is frequently understood to be a boon for rebel groups, key examples suggest that governance might not be as beneficial as previously supposed, and might even undermine the rebel group militarily. Terpstra and Frerks (2017) note that certain governance interventions, particularly rebel efforts to liberate women, can be met with antagonism, resentment and local hostility. Likewise, Arjona (2015: 195–197) recounts how the Fuerzas Armadas Revolucinarias de Colombia’s (FARC) efforts to control specific villages led to local-level mobilization against the rebel group. Moreover, the Islamic State faced active, violent civilian resistance in its capital of Raqqa (Khalaf, 2015: 58–59), and when the Syrian Kurdish Yekîneyên Parastina Gel (YPG) also controlled Raqqa, it also faced predominantly non-violent civilian resentment and resistance (Davison, 2017; @3z0ooz [Abdalaziz Alhamza], 2018).

One of the most infamous examples of the negative consequences of rebel governance is the People’s Liberation Army (PLA) and the Chinese Communist Party (CCP) during the Chinese Civil War (Opper, 2018). In the early-to-mid-1930s, the PLA was able to control territory throughout the South of China, and began laying the foundations of a new, radical social order there. Yet in so doing, it also created the conditions for its own collapse and near demise. In fact,

widespread civilian dissatisfaction with the forcible implementation of a radical land reform policy espoused by the CCP’s leadership from 1933 to 1934 brought about a massive and overwhelming defection of civilians from the CCP to the [Kuomintag], bringing about a complete collapse of the Chinese Soviet Republic. (Opper, 2018: 49–50)

The establishment of radical sociopolitical and economic institutions incited civilian defections and antagonism so severe that it hastened the collapse of the PLA in the South, instigating the now-infamous Long March. In other areas of China, Mao’s land reform initiatives would continue to provoke backlash by the one-time landowners against the peasant beneficiaries and even reduced agricultural productivity. Although important to Mao’s overall political goals, the “costs [of land reform] were apparently not matched by comparable gains” (Pepper, 1999: 308). Years later, Mao Tse-Tung remarked to a leader of the Cuban Revolution, Ernesto “Che” Guevara that the PLA’s radical initiatives were some of its biggest mistakes (Wilson Center, 1960).

The case of the PLA—one of the most pre-eminent examples of rebel governance—demonstrates how the governance interventions of rebel groups can be militarily deleterious to the rebel organization. In this case, rather than win over supporters, the PLA’s implementation of governance encouraged defections that served to unravel the organization’s foothold in southern China, and ultimately resulted in a retreat that would temporarily devastate the organization’s capacity. What the PLA case suggests is that governance erodes the military strength of the rebel group, so only the strongest rebels can undertake governance activities.⁷

Military benefits of governance limited or conditional

A second reason for the mixed findings between rebel capacity and rebel governance is that rebel governance does in fact improve the capacity of rebel groups, but improvements to the rebels’ military strength are either weak or conditional. For instance, one potential mechanism is that distinct governance institutions have differential effects on rebel strength: some institutional forms of governance can be highly beneficial militarily, while other institutional forms can erode rebels’ military strength.⁸ Rebels undertake a variety of different forms of governance that range from protection and security, to even the construction of a hydroelectric power plant for electricity (Lintner, 1990: 81) or video halls (Lintner, 1990: 91). From this range of possible governance institutions, some types might indeed be popular, facilitate collaboration and ease the extraction of resources and recruits from civilians. One example of a popular institutional form is security provision by creating police or communal defense groups. Both Kalyvas (2006: 124) and Mampilly (2011: 54) note that one of the most powerful ways to secure civilian collaboration during wartime is to protect civilians from state incursions. Although basic, this fundamental form of governance is absolutely one of the most crucial in securing civilian collaboration, generating local support and ultimately recruiting civilians to fight for the rebel cause.

Yet protection is far from the only form of governance that rebel groups provide and not all governance interventions are as popular as security from state-led military attacks. Some rebel governance interventions are deeply unpopular, and can even incite civilians to push back against, if not violently resist, their implementation (see, for example, Arjona, 2015; Opper, 2018). One example of the mixed consequences of different governance interventions involves the Eritrean People’s Liberation Front (EPLF). The EPLF sought to introduce a broad portfolio of governance institutions, ranging from protection and classes about how to optimize food production to significantly changing gender roles, removing pre-existing power holders and implementing land reform (see, e.g. Pool, 2001: 106–128). The EPLF, however, was aware that not all of its governance interventions would be equally desirable to all civilians in all places where the group controlled territory (Connell, 1993: 64). To account for that fact, it strategically introduced services in a piecemeal fashion (Connell, 1993: 64). First, the EPLF introduced some of the most popular forms of governance—a police force, veterinary services and agricultural initiatives—so that it could go about establishing control “without deeply upsetting local traditions or demanding complex skills” (Connell, 1993: 64). As the EPLF became more entrenched in the population, it knowingly introduced increasingly unpopular governance institutions, including the displacement of traditional authorities as well as investments in women’s education and land reform (Pool, 2001: 108–110, 127–128). Women’s education was so unpopular that female cadres were tasked with going to the homes of women to personally meet with and educate them in the hopes of “breaking taboos against female education” (Pool, 2001: 128).⁹ Traditional authorities at risk of losing their land to redistribution efforts spearheaded by the EPLF even turned to a rival Eritrean rebel group to violently cow civilians into refusing to implement any redistribution schemes (Houtart, 1980: 103).

In this case, resistance to unpopular governance policies meant that the EPLF had to divert additional resources to quell the violence within the territories that it held. The EPLF’s inability to protect civilians from violent reprisals against the very policies the group was attempting to implement could have eroded civilian trust and collaboration, potentially harming later attempts at extracting resources and recruiting followers. The mixed responses to different rebel institutions provided by the same rebel group in the same area could be the reason for the null findings between rebel governance and rebel strength. Stated otherwise, some rebel governance institutions are linked to enhancing rebels’ military capacity, but other governance interventions actively erode the rebels’ strength, producing a weak relationship between rebel strength and rebel governance.

A second potential mechanism is that rebel governance does in fact confer military benefits to rebel groups in terms of resources and personnel, but is financially costly and demands a large number of people to implement.¹⁰ These costs absorb much of any benefits that rebels might glean from governance, and the balance of both costs and gains accounts for the mixed and inconsistent association with rebel governance and rebel strength. For instance, in captured letters, a leader of al-Qaeda in the Arabian Peninsula admits that “control of these areas during one year cost us 500 martyrs, 700 wounded, 10 cases of hand or leg amputation and nearly $20 million. Most of the battle costs, if not all, were paid for through the spoils” (Callimachi, 2013). Likewise, the null relationship between rebel healthcare and rebel capacity observed in Tables 3 and 4 could be explained by the costliness of medicines, doctors, facilities, etc. Additionally, the costs of rebel governance could grow (both in terms of personnel and resources) as counterinsurgency pressures increase, further diminishing any positive consequences gleaned through governance, and draining the strength of the rebel group (Revkin and McCants, 2015; Revkin, in press).

Relatedly, a third mechanism is that the effects of rebel governance on military capacity are conditional on pre-existing operational factors, such as the quality of pre-existing state institutions (Arjona, 2015, 2016; Ahmad, 2017; Mukhopadhyay, 2014), or pre-war networks of local support (Staniland, 2014).¹¹ In places where pre-existing institutions are weak or where rebels enjoy high levels of pre-existing support, rebel governance affirms civilians’ beliefs in both the rebellion and rebels’ capacity to achieve victory over the government (Grynkewich, 2008; Mampilly, 2011). Under these conditions, the introduction of governance may increase rebel support in ways that facilitate recruitment and the collection of resources.

Yet the same rebel group may confront more hostile populations where rebel governance initiatives are met with skepticism at best, if not outright hostility (Arjona, 2015), in places where state institutions are strong or local support is weak. The contrasting civilian reactions to the same rebel governance interventions could explain the mixed results for the military consequences of rebel governance: rebel governance yields military benefits in some areas, but is met with hostility and resistance in other areas, eliminating any military benefits from governance that rebels might enjoy.

One example of this dynamic is the FARC. Arjona (2016) explains how the FARC’s governance interventions in Tellus and Librea were profound and touched upon many aspects of civilians’ personal and public lives (Arjona, 2016: 239–244). Here, state institutions were relatively weak, allowing rebels to more easily consolidate their rule (Arjona, 2016: 233). The civilians’ response to these initiatives was largely positive: “Cooperation—in the form of both obedience and spontaneous support—was widespread. Most people obeyed the myriad rules established by the group. … Many sought ways to please combatants in order to be on good terms with them. … Although cooperation was widespread, adolescents were particularly fond of the FARC. To them, combatants were ‘someone to emulate, people with power, an option to imitate’” (Arjona, 2016: 244).

In contrast, the very same rebel group encountered intense resistance when attempting to introduce the same portfolio of institutions in different places where local institutions were far more robust (Arjona, 2015: 195–197). Where local institutions were strong in Cauca, Colombia, locals resisted the FARC’s attempts to consolidate order, organizing local protection groups and sometimes even physically removing the FARC (Arjona, 2015: 195–197). Despite the FARC’s efforts to control a strategically important region (Arjona, 2015: 196), civilians organized to resist its efforts.

A fourth and final explanation is that, rather than confer any military benefits to the rebel group, governance may simply appease civilians enough so that they accept rebels’ dominion over them. Governance does not result in military benefits for the rebel group, rather, it prevents the provocation of civilian resistance against rebels out-migrating from rebel-held territories or civilian defections to and collaboration with the state. In this sense, the positive consequences of rebel governance are absences: the absence of civilian resistance, the absence of defectors, and the absence of mass flight. Instead of gaining in military strength from governance, rebels simply do not grow any weaker from civilian actions.

Revkin (2019: 4) demonstrates that while some of the Islamic State’s governance interventions were popular, not all residents of Mosul were equally supportive of its behavior. Instead, Revkin (2019: 3, 6–7) finds that Mosul residents stayed because the Islamic State’s governance was actually relatively good compared with the Iraqi government’s previous provision of services. The residents of Mosul were willing to accept repressive policies and stay within the city simply because governance was better (Revkin, 2019: 10). Nevertheless, many Mosul residents still did not accept the group’s ideology, were personally victimized by rebels, and complied with rebel commands out of fear for their lives (Revkin, 2019: 4). Rather than emboldening civilians to join the rebels’ cause, the Islamic State’s governance prevented people from fleeing their aspiring caliphate.

In this section, I have used qualitative vignettes to suggest four potential reasons why governance and military capacity might not share a positive association. One explanation is that governance has conditionally positive consequences: only some rebel governance initiatives have positive consequences, while others actively erode rebels’ capacity. A second explanation is that governance has certain military benefits but is also extremely costly, so increases in rebel group expenses could mean that the benefits of governance do not outweigh the costs. A third, similar explanation is that governance’s militarily beneficial consequences could be conditional on pre-existing local conditions. Fourth and finally, it is possible that rebel governance has no military benefits, it simply prevents defections, out-migration or civilian resistance.

Causal link between governance and strength misspecified

A third set of explanations that could account for the absent or negative relationship between rebel governance and rebel military capacity is that the causal arrow between military capacity and governance has been misspecified. One way to interpret the null findings of the relationship between rebel governance and rebel military capacity is that the weakest rebels use governance, while the strongest rebel groups have no need to do so. Therefore, the hypothesized causal direction between rebel governance and rebel strength (that rebel governance yields military benefits, thereby increasing the strength of the rebel group) is actually reversed.

In this case, rebel governance does in fact have military benefits and confers resources and recruits upon rebel groups. However, rebels only govern when they lack weapons and personnel, and thus they are weak. Because governance is only employed by weaker rebels, the relationship between governance and strength appears ambiguous if not negative. Such an explanation is broadly consistent with works by Weinstein (2006) and Huang (2016). Both scholars argue that rebels turn to governance to fill organizational or resource holes. For instance, Weinstein (2006) argues that when rebels have more social endowments than economic ones, one way rebels can recruit civilians is to rely on governance or mutually beneficial processes of exchange with civilians. Similarly, Huang (2016) argues that when rebels are more reliant on civilians for support, they tend to provide governance to better facilitate exchange between the two groups.¹²

An alternative explanation is that rebel governance and rebel military capacity are wholly unrelated—that the hypothesized causal link simply does not exist. Factors aside from governance initiatives can drive both recruitment and access to resources, and therefore rebel strength. Instead, rebels can use their governance for ideological, performative or goals-oriented reasons directed at invested onlookers within and outside the conflict (Mampilly, 2011, 2015; Revkin, in press; Stewart, 2018). For instance, US State Department archival documents show that the Eritrean Liberation Front’s (ELF) recruitment process was relatively straightforward and completely unrelated to governance activities. The ELF’s recruitment could largely be explained by:

(A) political or ideological commitment to its goals, which accounts for a very small percentage; (B) in Moslem and tribal areas, a feeling by young that they should “prove their manhood” by “becoming fighters”; an ELF recruiter will come into a village and taunt young farmers or herdsmen, saying “Mohamed has been fighting with us for a year already, and you, Omar, remain behind in the village with the women and children. Mohamed is a man and your are nothing but an old woman”; (C) fear of [Ethiopian] army sweeps. (US State Department, 1969: 2)

None of these extraction processes were in any way related to the ELF’s governance, which the group claimed included some forms of education and healthcare by the mid-1970s (Eritrean Liberation Front 1977).

The ELF demonstrates that recruitment need not involve governance. Thus, the lack of a statistically significant correlation between governance and military capacity is because these two processes are wholly unrelated, serve two different functions and achieve two different outcomes. Rather than use governance to draw strength from the civilian population, the ELF used coercion, promises and psychological appeals, while also harnessing the fallout of a brutal government counterinsurgency campaign.

The cases above suggest that the lack of a statistical relationship between governance and military capacity could be explained by a misspecified causal process, in that the relationship is endogenous or that no causal link between governance and military capacity exists. If the relationship between governance and capacity is endogenous, then weak rebels could systematically turn to governance to supplement their weakness. Alternatively, the ELF case shows that governance and recruitment were largely unrelated to one another: the ELF recruited and extracted through psychological appeals and promises, by capitalizing on state repression or by coercion.

Analytic and data limitations

One final reason for the surprising findings about rebel governance and rebel strength is that the null and negative findings are a product of data limitations arising from high levels of aggregation over space and time (rebel-group unit of analysis), or from inadequate conceptualizations and operationalizations of rebel governance and rebel strength.

For instance, the analysis above focused on education and healthcare, which are not necessarily the most economic social services, nor are they necessarily the most likely to generate the greatest possible benefits for rebel mlitaries. Furthermore, by examining provision to people who are unlikely to join the rebel movement (Table 4), I am inherently examining more costly forms of governance to understand if these costly investments yield greater dividends. Data on more cost-effective institutions like taxation (see Revkin, in press as an example of subnational data collection) as well as unpopular institutions like land reform or women’s liberation initiatives must be analyzed alongside data about more popular forms of rebel institutions, such as policing, healthcare or agricultural services, which could shed light on the heterogeneous consequences of rebel governance. Already, scholars like Huang (2016), Florea (2014), Heger and Jung (2017) and Stewart (2018) have begun collecting data that disaggregate by institutions, but many are missing some of the governance interventions highlighted above that frequently provoke antagonism. Likewise, subnational rebel governance data collection could identify location-dependent variation in the consequences of rebel governance. Both Revkin (in press) and Arjona (2016) are excellent examples of pioneering work on subnational-level rebel governance data collection.

In terms of rebel capacity, this discussion reveals the need for scholars to take seriously the myriad ways rebel capacity could manifest and be codified. Rebel military capacity could take the form of technical expertise, weaponry, military and/or financial aid and/or the size of its ranks and reservists. Rebel strength could also reflect the coercive capacity of the organization. For instance, the inclusion of rebels’ capacity to implement and maintain deeply coercive and unpopular governance policies that imperil combatants might in fact be one indication of rebel strength. Additionally, data on rebel strength are also time-invariant, but as the opening anecdote illustrates, the relationship between rebel strength and rebel governance can also vary and change over time (see also Revkin, in press).¹³ Ultimately, this finer conceptualization of rebel capacity, or the multiple dimensions that encompass it, ought to be complemented by richer and more precise time-variant and subnational quantitative data collection, much like measures of rebel governance. All told, a more precise and developed codification of rebel strength might also reveal that rebel governance and rebel strength are more closely and positively related to each other.¹⁴

Summary and discussion

Using primary and secondary sources, I suggest four broad categories of explanations that could explain the never positive, but sometimes statistically significantly negative relationship between rebel strength and rebel governance. One category of explanation rejects the assumption that rebel governance yields military benefits for the rebel group. Instead, governance weakens rebels’ organizational capacity, sometimes to their extreme peril. A second category of explanations conditionally accepts the positive consequences of rebel governance, but identifies limits to these benefits: upfront costs, uneven civilian appreciation of governance interventions and varied responses to certain institutions. A third category of explanations suggests that the causal arrow linking governance and capacity is misspecified. Finally, the last category of explanation details how this relationship could be the result of data limitations. The purpose of this discussion is to put forth a set of empirically grounded explanations that are neither mutually exclusive nor competing and that illuminate why rebel governance lacks a positive relationship, let alone a statistically significant one, with rebel strength. In so doing, this article is not in any way intended to suggest that research premised on the assumption that rebel governance has militarily beneficial consequences is incorrect. Rather, the purpose of the discussion above is to encourage scholars to identify the conditions under which rebel governance improves rebels’ military capacity, and under which conditions rebel governance does not.

Conclusion and implications

In this paper, I investigated one hypothesized consequence of rebel governance: increased rebel group capacity. One of the most prominent theoretical explanations for why rebels govern in the literature today is that rebel governance generates popular support, and this popular support facilitates rebel recruitment and the extraction of resources. With more recruits and resources, rebels are stronger and better able to contest the state militarily. Relying on an original dataset, I conducted a preliminary and non-causal statistical analysis to evaluate the strength of the relationship between rebel governance and rebel strength. What I found was surprising: rebel governance at best has no relationship to rebel strength, and at worst is negatively associated with rebel strength. To account for this surprising finding, I turned to primary and secondary qualitative evidence to inductively generate a set of possible explanations that could explain the null or negative results.

In so doing, this paper makes a number of contributions. First, it offers a preliminary and ultimately descriptive test of some of the assumptions regarding the consequences of rebel governance. Second, it offers a number of potential mechanisms that could explain why rebel governance does not appear to enhance rebel military capacity. Third, by questioning the military utility of governance, I highlight the need for greater theorization about the strategic motivations underlying rebels’ choice to govern, as well as why some rebels elect to govern in ways that they know could be unpopular.

This paper also highlights several avenues for future work: both theoretical and empirical. Empirically, this paper calls for a richer investment in quantitative and qualitative data collection, both subnationally and cross-nationally, in ways that also account for temporal variation. Such data-collection efforts enable scholars not only to begin testing some of the fundamental assumptions that underlie this body of research, but also to untangle the consequences of rebel governance during conflict. For instance, quantitative data collection on a far more vast array of governance institutions could enable scholars to identify whether certain types of governance services serve different functions or have different consequences among different populations at different periods of time during war. Likewise, subnational data collection on rebel governance could help isolate the micro-level consequences of rebel governance that might not generalize or aggregate up to higher levels of analysis. Finally, scholars could also begin to conceptualize and collect measures of data quality to test whether it is not simply the presence of rebel institutions but the quality thereof that is most tightly related to rebel strength. Such data collection for quantitative research cannot of course be accomplished without investing in rigorous qualitative scholarship.

Theoretically, noting that rebels could undertake governance to reap benefits outside of a straightforward improvement in its strength opens up a range of potential reasons for why certain rebel groups might undertake governance. Because rebel governance does not appear to have consistently positive military consequences for the rebel group, researchers ought to question the strategic motivations undergirding some rebel groups’ choices to engage in governance at all. If some governance interventions are not beneficial and can be militarily detrimental, why do rebels pursue such governance behaviors? Is rebels’ choice to implement an unpopular policy a miscalculation? Perhaps more puzzlingly, do rebels introduce policies that they know might be unpopular? If so, these surprising behaviors could suggest that deeply rooted ideological (Mampilly, 2011; Revkin, in press) or goal-based commitments (Stewart, 2018) may motivate certain rebel groups to provide more governance, as opposed to military calculations about increasing rebels’ strength. As a result, researchers moving forward ought to consider rebel strategies for the provision of any governance (and variation in the extent rebels seek govern), as well as the structural, organizational or wartime factors that impede or accelerate the implementation of governance in the first place.

Supplemental Material

cmp881422_supplmater – Supplemental material for Rebel governance: military boon or military bust? (Isard Award Article)

Supplemental material, cmp881422_supplmater for Rebel governance: military boon or military bust? (Isard Award Article) by Megan A Stewart in Conflict Management and Peace Science

Footnotes

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Megan A. Stewart

Supplemental material

The Online Appendix is available online through the SAGE CMPS website:

Notes

References

@3z0ooz [Abdalaziz Alhamza] (2018, 3 June) A thread about what’s going on in#Raqqa right now 1-on May 27th a protest hit the streets of Raqqa calling on #YPG/#SDF to leave #Raqqa city because of the atrocities that they have been committed against the civilians and the forcing recruitment for even the children. Twitter. Available from: https://twitter.com/3z0ooz/status/1003372074780577794 (accessed 4 June 2018).

Ahmad

(2017) Jihad & Co.: Black markets and Islamist Power. Oxford: Oxford University Press.

Amrane

Abu-Haidar

(1999) Women and politics in Algeria from the War of Independence to our day. Research in African Literatures 30(3): 62–77.

Anderson

(1978) Inside Sandinista camp: Working to overthrow Nicaragua’s Somoza. The Saint Petersburg Times, Letters Section 15A. Available from: https://news.google.com/newspapers?nid=feST4K8J0scC&dat=19781115&printsec=frontpage&hl=en, p. 13 (accessed 2015).

Arjona

(2015) Civilian resistance to rebel governance. In: Arjona

Mampilly

Kasfir

(eds), Rebel Governance in Civil War. New York: Cambridge University Press, pp. 180–202.

Arjona

(2016) Rebelocracy: Social Order in the Colombian Civil War. Cambridge: Cambridge University Press.

Balcells

(2017) Rivalry and Revenge. Cambridge: Cambridge University Press.

Balcells

(2018) Dynamics of internal resettlement during civil war: Evidence from Catalonia (1936–39). Journal of Peace Research 55(2): 236–251.

Balcells

Kalyvas

(2015) Revolutionary rebels and the Marxist Paradox. Work in Progress. Available from: http://cpd.berkeley.edu/wp-content/uploads/2015/04/MarxIns_4_15.pdf (accessed 30 May 2017).

10.

Callimachi

(2013, 9 August) Yemen terror boss left blueprint for waging Jihad. Associate Press Available from: https://www.pulitzer.org/finalists/rukmini-callimachi-0 (accessed 29 January 2019).

11.

CAVR (2005) Part 5: Resistance: Structure and strategy. In: Chega! Final Report of the Commission for Reception, Truth and Reconciliation in East Timor (CAVR). Commission for Reception, Truth, and Reconciliation Timor–Leste. Available from: https://www.etan.org/etanpdf/2006/CAVR/05-Resistance-Structure-and-Strategy.pdf (accessed 25 October 2019).

12.

Connell

(1993) Against All Odds: A Chronicle of the Eritrean Revolution: With a New Foreword on the Postwar Transition. Trenton, NJ: The Red Sea Press.

13.

Cunningham

Gleditsch

Salehyan

(2009) It takes two: A dyadic analysis of civil war duration and outcome. Journal of Conflict Resolution 53(4): 570–597.

14.

Davison

(2017, 7 September) Some Syrian schools erase Assad but tensions rise over Kurdish. Reuters. Available from: https://www.reuters.com/article/us-mideast-crisis-syria-kurds/some-syrian-schools-erase-assad-but-tensions-rise-over-kurdish-idUSKCN1BI1X6 accessed (23 May 2018).

15.

Dhada

(1993). Warriors At Work: How Guinea Was Really Set Free. Niwot, CO: University of Oklahoma Press, 1993.

16.

Eck

(2018) The origins of policing institutions: Legacies of colonial insurgency. Journal of Peace Research 55(2): 147–160.

17.

Eritrean Liberation Front (1977) Eritrean Liberation Front: 16 Years of Armed Struggle. Beirut: The Eritrean Liberation Front.

18.

Fearon

Laitin

(2003) Ethnicity, insurgency, and civil war. American Political Science Review 97(1): 75–90.

19.

Florea

(2014) De facto states in international politics (1945–2011): A new data set. International Interactions 40(5): 788–811.

20.

Green

(2018) The Commander’s Dilemma: Violence and Restraint in Wartime. Ithaca, NY: Cornell University Press.

21.

Grynkewich

(2008) Welfare as warfare: How violent non-state groups use social services to attack the state. Studies in Conflict & Terrorism 31(4): 350–370.

22.

Hassan

O’Mealia

(2018) Uneven accountability in the wake of political violence: Evidence from Kenya’s ashes and archives. Journal of Peace Research 55(2): 161–174.

23.

Heger

Jung

(2017) Negotiating with rebels: The effect of rebel service provision on conflict negotiations. Journal of Conflict Resolution 61(6): 1203–1229.

24.

Herbst

(2014) States and Power in Africa: Comparative Lessons in Authority and Control. Princeton, NJ: Princeton University Press.

25.

Heston

Summers

Aten

(2012) Penn World Table Version 7.1 Dataset. Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania.

26.

Högbladh

Pettersson

Themnér

(2011) External support in armed conflict 1975–2009: Presenting new data. In: 52nd Annual International Studies Association Convention, Montreal, pp. 16–19.

27.

Houtart

(1980) The social revolution in Eritrea. In: Davidson

Cliffe

Selassie

(eds) Behind the War in Eritrea. Nottingham: Spokesman Books, pp. 83–110.

28.

Huang

(2016) The Wartime Origins of Democratization: Civil War, Rebel Governance, and Political Regimes. Cambridge: Cambridge University Press.

29.

Human Rights Watch/Africa and Human Rights Watch Children’s Rights Project (1997) The Scars of Death: Children Abducted by the Lord’s Resistance Army in Uganda, Vol. 3169. Human Rights Watch.

30.

Jackson

(2018) Life under the Taliban shadow government. Overseas Development Institute. Available from: https://www.odi.org/publications/11144-life-under-taliban-shadow-government (accessed 22 January 2018).

31.

Kalyvas

(2006) The Logic of Violence in Civil War. Cambridge: Cambridge University Press.

32.

Kalyvas

(2018) Jihadi rebels in civil war. Dædalus 147(1): 36–47.

33.

Kasfir

(2005) Guerrillas and civilian participation: The National Resistance Army in Uganda, 1981–86. The Journal of Modern African Studies 43(2): 271–296.

34.

Khalaf

(2015) Governance without government in Syria: Civil society and state building during conflict. Syria Studies 7(3): 37–72.

35.

Levi

(1989) Of Rule and Revenue. Berkeley, CA: University of California Press.

36.

Lidow

(2016) Violent Order: Understanding Rebel Governance Through Liberia’s Civil War. Cambridge: Cambridge University Press.

37.

Lintner

(1990) The Rise and Fall of the Communist Party of Burma (CPB). Ithaca, NY: Cornell South East Asia Program.

38.

Mahanta

(2013) Confronting the State: ULFA’s Quest for Sovereignty. New Delhi: SAGE.

39.

Mamdani

(2003) Making sense of political violence in postcolonial Africa. In: War And Peace in the 20th Century And Beyond. Singapore: World Scientific, pp. 71–99.

40.

Mampilly

(2011) Rebel Rulers: Insurgent Governance and Civilian Life During War. Ithaca, NY: Cornell University Press.

41.

Mampilly

(2015) Performing the nation state: Rebel governance and symbolic processes. In: Arjona

Kasfir

Mampilly

(eds) Rebel Governance in Civil War. New York: Cambridge: Cambridge University Press.

42.

Menkhaus

(2007) Governance without government in Somalia: Spoilers, state building, and the politics of coping. International Security 31(3): 74–106.

43.

Mukhopadhyay

(2014) Warlords, Strongman Governors, and the State in Afghanistan. Cambridge: Cambridge University Press.

44.

Olson

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

45.

Opper

(2018) Revolution defeated: The collapse of the Chinese Soviet Republic. Twentieth-Century China 43(1): 45–66.

46.

Osorio

Schubiger

Weintraub

(2018) Disappearing dissent? Repression and state consolidation in Mexico. Journal of Peace Research 55(2): 252–266.

47.

Paris

(2004) At War’s End: Building Peace after Civil Conflict. Cambridge: Cambridge University Press.

48.

Parkinson

Zaks

(2018) Militant and rebel organization(s). Comparative Politics 50(2): 271–293.

49.

Pepper

(1999) Civil War in China: the Political Struggle 1945–1949. Lanham, MD: Rowman & Littlefield.

50.

Pool

(2001) From Guerrillas to Government: the Eritrean People’s Liberation Front. Athens, GA: James Currey.

51.

Revkin

(in press) What explains taxation by resource-rich rebels? Evidence from the Islamic State in Syria. Journal of Politics. Available from: https://www.journals.uchicago.edu/doi/abs/10.1086/706597?journalCode=jop

52.

Revkin

(2019, 14 October) Relative legitimacy and displacement decisions during rebel governance. Available from SSRN: https://ssrn.com/abstract=3365503 or https://dx-doi-org.web.bisu.edu.cn/10.2139/ssrn.3365503

53.

Revkin

McCants

(2015, 20 November) Experts weigh in: Is ISIS good at governing? Brookings Institute. Available from: https://www.brookings.edu/blog/markaz/2015/11/20/experts-weigh-in-is-isis-good-at-governing/ (accessed 29 January 2019).

54.

Scharpf

(2018) Ideology and state terror: How officer beliefs shaped repression during Argentina’s “Dirty War”. Journal of Peace Research 55(2): 206–221.

55.

Schwartz

Straus

(2018) What drives violence against civilians in civil war? Evidence from Guatemala’s conflict archives. Journal of Peace Research 55(2): 222–235.

56.

Staniland

(2014) Networks of Rebellion: Explaining Insurgent Cohesion and Collapse. Ithaca, NY: Cornell University Press.

57.

Stearns

(2012) Dancing in the Glory of Monsters: The Collapse of the Congo and the Great War of Africa. New York: Public Affairs.

58.

Stewart

(2018) Civil war as state-making: Strategic governance in civil war. International Organization 72(1): 205–226.

59.

Stewart

Kitchens

(2018) Statebuilding, social order and violence: Evidence from US reconstruction. Available from: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3223825 .

60.

Stewart

Liou

Y-M

(2017) Do good borders make good rebels? Territorial control and civilian casualties. The Journal of Politics 79(1): 284–301.

61.

Sullivan

Davenport

(2018) Resistance is mobile: Dynamics of repression, challenger adaptation, and surveillance in US “Red Squad”and black nationalist archives. Journal of Peace Research 55(2): 175–189.

62.

Terpstra

Frerks

(2017) Rebel governance and legitimacy: Understanding the impact of rebel legitimation on civilian compliance with the LTTE Rule. Civil Wars 19(3): 279–307.

63.

US State Department (1969) A Moslem view of the ELF problem. Research Group 59, 1969. Airgram, file POL 23-9 ETH, 1967–1969 Subject-Numeric File, RG 59: General Records of the Department of State, US National Archives, 30 June 1969. Adelphi, MD (accessed 18 July 2018).

64.

Weinstein

(2006) Inside Rebellion: The Politics of Insurgent Violence. Cambridge: Cambridge University Press.

65.

Wilson Center (1960) Memorandum of conversation between Mao Zedong and Ernesto “Che” Guevara. Translation of memorandum from November 19, 1960. PRC FMA 202-00098-01, 1-14. Trans. by Zhang

Qian

. Available from: http://digitalarchive.wilsoncenter.org/document/115155.

66.

World Bank (2012) World Development Indicators 2012. New York: World Bank Publications.

67.

Zhukov

Talibova

(2018) Stalin’s terror and the long-term political effects of mass repression. Journal of Peace Research 55(2): 267–283.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB