A primary goal of international aid in conflict-affected countries is to help political actors build public support by winning the ‘hearts and minds’ of the civilian population. However, it remains unclear whether – and under what conditions – aid can achieve these objectives. Most studies of foreign aid in war zones have focused on assistance provided to states, often as part of counterinsurgency operations. This research has yielded mixed results regarding the impact of aid on governance outcomes.Footnote 1 In this article, we examine the effects of external assistance designed to bolster the authority of rebel governors, who have also been recipients of aid in conflict-ridden countries.
Building on insights from the literatures on civil war and state formation, we theorize that foreign aid's ability to enhance popular support for rebel governors critically depends on whether military control is contested between different armed actors. In uncontested areas, aid helps rebel governors demonstrably deliver much-needed services to the population and garner public support as a result. In contested areas, we argue that aid cannot bolster opinions of opposition governing institutions for three reasons. First, rebel governors lose credibility if they cannot offer the most valuable public good during war: protection. Secondly, the delivery of efficacious, sustainable services is complicated, if not stymied, by competition from rivals. Thirdly, contestation creates a significant attribution challenge for institutions that seek credit for providing public goods.
To test our theory, we focus on non-military US aid provided during the Syrian civil war. This conflict is a protracted, internationalized struggle between multiple warring parties, and thus shares key features common to many modern armed conflicts. Moreover, the Syrian context allows us to assess the role of local institutions that developed as part of a popular civilian uprising, evolved into agents of governance, and became targets of humanitarian and development assistance from international donors. Since 2011, donors have provided non-military aid to members of the Syrian opposition in an effort to bolster ‘good’ rebel governors, funneling more than a billion dollars in ‘politically oriented assistance’ to local governing councils in rebel-held communities (Brown Reference Brown2018, 2). The Syrian case therefore enables us to examine the effects of this assistance on public perceptions of rebel governing institutions.
We utilize a novel dataset that combines information on aid provided by the US Agency for International Development's Office of Transition Initiatives (USAID/OTI) with civilian perceptions data collected from residents of twenty-seven opposition-held communities in Syria. The data include more than 13,000 surveys and 1,000 in-depth interviews, which provide a unique opportunity to examine public opinion in an area that has been virtually inaccessible to researchers. We combine this data with detailed, community-level aid data to test our theory both quantitatively and qualitatively. We supplement these data sources with information gathered during fieldwork that we conducted in Syria along its border with Turkey between 2013 and 2019.
Our analysis yields several results. First, we find that in communities unmarked by intra-opposition conflict, aid provided by USAID/OTI had a positive effect on public support for local opposition governing councils. However, it did not boost support for councils in communities experiencing contestation (see Tao et al. Reference Tao2016) – when the presence of one rebel group was challenged by at least one other rebel group. These results hold in a variety of robustness checks, including an instrumental variables specification to address potential selection issues. We also find a similar result at the national level of governance: contestation between the Syrian regime and rebel forces conditioned the impact of aid on public perceptions of both the councils and the national opposition body, the Syrian Interim Government. Finally, our qualitative analysis of opposition politics in Aleppo City and Saraqeb allows us to trace the mechanisms by which violent contestation can condition the influence of aid on public support for institutions that aspire to govern during conflict.
This study makes three main contributions. First, we examine the multiple dimensions of competition over violence that can characterize the thorniest of civil wars and their implications for governance and intervention. In doing so, we advance the burgeoning literature on the relationship between foreign aid and governing authority. There is no consensus within this body of research (reviewed below) over whether aid-funded service delivery can bolster legitimacy – defined here as ‘social acceptance of the right to rule’ (Risse and Stollenwerk Reference Risse and Stollenwerk2018). We argue that the effect is conditional: for non-military aid to have a positive effect on popular perceptions of governing institutions, we theorize that it must be provided in uncontested spaces. Moreover, we demonstrate the precise mechanisms driving this effect. Secondly, the article underscores the oft-obscured distinction between civilian and military institutions in the work of insurgent rule, demonstrating the importance of each and their relationship to one another for counter-state-building. Finally, we contribute to policy-oriented debates about the utility of attempts to ‘win hearts and minds’ and advance governance aims in conflict-affected environments, as our results suggest that territorial control should be seen as a prerequisite for effective foreign assistance in war zones.
Theory
Despite the enormous amount of development aid directed toward conflict-affected countries,Footnote 2 prior research is ambivalent on the ability of foreign assistance to ‘win hearts and minds’. While some studies of foreign aid in general have revealed a positive relationship between aid and public support for governing institutions, many find little evidence of sustained political change. Some scholars argue that aid erodes the ‘fiscal contract’ in which states and citizens trade taxation for the accountable provision of public goods (Bates and Donald Lien Reference Bates and Donald Lien1985; Timmons Reference Timmons2005). While aid can provide much-needed services (Böhnke and Zürcher Reference Böhnke and Zürcher2013; Gupta Reference Gupta1999), it may also weaken states' credibility when governments make unfulfilled claims about their capacity to provide for their constituents (Gubser Reference Gubser2002; Lake Reference Lake2010).Footnote 3 Additionally, citizens may detect favoritism, inefficiency or corruption in the distribution of aid (Ahmed Reference Ahmed2012; Böhnke and Zürcher Reference Böhnke and Zürcher2013; Brass Reference Brass2016; Fishstein and Wilder Reference Fishstein and Wilder2012; Jablonski Reference Jablonski2014).Footnote 4 Moreover, donor-branded aid is often attributed to foreign actors rather than their local clientele, though governments sometimes succeed at claiming credit for it (Cruz and Schneider Reference Cruz and Schneider2017; Guiteras and Mobarak Reference Guiteras and Mobarak2015).Footnote 5 Indeed, many empirical studies have found either a negative or null relationship between aid and legitimacy, particularly in conflict settings (Beath, Christia and Enikolopov Reference Beath, Christia and Enikolopov2015; Humphreys, de la Sierra and Van der Windt Reference Humphreys, de la Sierra and Van der Windt2015; King and Samii Reference King and Samii2014; Nixon and Mallett Reference Nixon and Mallett2017; Nunn and Qian Reference Nunn and Qian2014).Footnote 6
While foreign aid can influence citizens' perceptions of governance in multiple ways, the primary pathway explored in previous research – and the one on which most donor programs are predicated – is its ability to boost support by helping to increase service provision. Focusing on this theory of change, we argue that there is more room for optimism than the existing literature suggests when considering the effects of aid on public support for rebel governors, which are rebel groups that take up the project of governing.Footnote 7
We theorize that aid can boost the opinions of these actors for at least three reasons. First, the delivery of basic services is often an exceptionally high priority for populations living in conflict-affected territories that are no longer privy to the government's distribution of public goods. By helping rebel governors provide these services, aid may enhance their support among the public and help rebels signal their viability as a governing alternative to the state. In many instances, aid is one of the only ways that rebels can fund service delivery and other core governmental functions, and the public is unlikely to expect otherwise given the limited extractive capabilities and difficulties associated with taxing war-affected populations that most insurgencies face.Footnote 8 Secondly, rebels frequently govern sympathetic communities whose disgruntlement with their government served as the very basis for the insurgency. As a result, these populations may be more inclined to support aid-facilitated efforts to build an alternative to the political status quo. Finally, because counter-states lack, and therefore seek, the juridical sovereignty that states enjoy, their constituents may view the successful receipt of aid not as a sign of weakness, but rather as an achievement in the form of international recognition (Bob Reference Bob2005; Coggins Reference Coggins, Pavković and Radan2011).Footnote 9 Thus, in our context, we anticipate that aid can enhance public support.
However, drawing on lessons from the state formation and civil war literatures, we expect any positive effect of aid to be conditioned by the level of contestation in areas where it is administered. In contested areas, it is difficult for rebel governors to use aid to win ‘hearts and minds’ for three key reasons. First, governing bodies in such areas cannot rely solely on service provision to improve their credibility. While aid may help governors provide better services, people are primarily concerned with their security in the face of violent conflict. Hard forms of assistance – military and otherwise – are required under such circumstances; other kinds of assistance may be helpful, but their import pales in comparison. Control is often described as the sine qua non for insurgent success, since it ‘signals credibility’ and enables conflict parties to deploy a wide range of measures – including aid – that facilitate the civilian collaboration necessary to advance their cause (Kalyvas Reference Kalyvas2006, 114). The civil war literature recognizes insurgency as a kind of ‘competitive state-building’ and stresses the paramount importance of territorial control over violence for those on the rebel side. Indeed, the state formation literature has also long conceived of government as requiring concentrated control over violence. Meanwhile, the counterinsurgent motto of ‘clear, hold, and build’ suggests the need for control, even as in practice, governments and donors tend to employ aid ‘to help establish control of contested areas’ (Sexton (Reference Sexton2016), emphasis added).Footnote 10
A second and related reason we do not expect aid to improve public support in contested areas is because contestation undermines the efficacy of aid-enabled services, particularly when they are delivered by civilian institutions that lack the coercive capacity to protect those services in a conflict-affected environment. For example, an aid-funded electrical grid may be destroyed, or a water tank blown up or contaminated (Crost, Felter and Johnston Reference Crost, Felter and Johnston2014). In contrast, when armed groups provide services directly, they tend to defend them, making aid provided by civilian governing bodies look especially unreliable by comparison.Footnote 11 In contested areas, aid may also be siphoned off by competing actors. In uncontested spaces, we expect armed actors to avoid disrupting services and to even lend their support to civilian opposition institutions engaged in public goods provision. Many rebels perceive a political pay-off from governance projects and recognize the value of service provision. As Mampilly (Reference Mampilly2012, 54) argues, ‘since civilians in rebel-controlled areas can and do enjoy these goods without directly participating in the insurgency, the provision of public goods can be viewed as part of a broader program to generate legitimacy and support for the rebel regime’.
The third reason it is difficult for rebel governors to win ‘hearts and minds’ using aid is that attribution challenges that can arise with all forms of foreign aid grow more acute in violent contexts, making it especially difficult to establish claims as an authoritative governor. The ‘correct attribution of governance success and failure’ is a crucial aspect of translating governing effectiveness into political legitimacy (Risse and Stollenwerk Reference Risse and Stollenwerk2018, 413). When multiple groups share space in uncoordinated terms, local residents may struggle to accurately attribute the goods and services on offer to the entity that supplied them. In contested communities in the Syrian province of Idlib, for example, administrative bodies run by Salafi Islamist rebel groups Jabhat al-Nusra and Ahrar al-Sham took credit for services provided by US-funded civilian local councils aligned with factions of the Free Syrian Army (FSA), diverting reputational benefits away from these councils.Footnote 12 Given evidence that populations have ambiguous responses to branded aid (Winters, Dietrich and Mahmud Reference Winters, Dietrich and Mahmud2017), even equipment marked with a client institution's insignia may not prove adequately clarifying.Footnote 13 In the absence of armed contestation, we expect dominant armed groups to coalesce around, invest in, and even promote the work of their civilian counterparts in aid-enabled service delivery, precisely because of the presumed political dividends that accrue to effective rebel governors.Footnote 14 A coherent approach on the part of military and civilian actors is likely to be read more easily and accurately by civilian observers.Footnote 15
We specifically consider how these mechanisms operated during the Syrian civil war, because it offers a unique opportunity to consider the effect of aid on support for governing bodies affiliated with an aspiring counter-state. The Syrian context involves a significant foreign campaign to support the establishment of ‘good’ rebel governance, as well as a fragmented warscape with a range of aspiring rebel governors. Indeed, the Syrian insurgency featured a ‘lack of exclusive control of territories by its armed groups’ (Baczko, Dorronsoro and Quesnay Reference Baczko, Dorronsoro and Quesnay2018, 37); as in other civil wars, state and counter-state forces routinely coexisted in a variety of configurations over time and space (Staniland Reference Staniland2012). Rebel forces not only shared space with the state (and its foreign supporters) but also with one another in ways that often precluded consolidation for all involved parties (Bakke, Cunningham and Seymour Reference Bakke, Cunningham and Seymour2012; Fjelde and Nilsson Reference Fjelde and Nilsson2012; Pischedda Reference Pischedda2018; Wood and Kathman Reference Wood and Kathman2015). Multiple factions therefore wielded violence and, in many cases, fought, lived and governed in close proximity.
Though this provides an ideal setting in which to test our theory, we note that Syria is far from unique in this regard. Similar multi-faceted, internationalized civil wars featuring considerable rebel fragmentation continue in many countries today – including Afghanistan, the Democratic Republic of Congo, Iraq, Libya, Ukraine and Yemen (Fjelde and Nilsson Reference Fjelde and Nilsson2012; Fjelde and Nilsson Reference Fjelde and Nilsson2018). Indeed, a number of rebel groups take on the work of governing civilians who live in their midst. Advanced forms of rebel governance often include what Kasfir (Reference Kasfir, Arjona, Kasfir and Mampilly2015, 21) calls ‘civilian administration’, whereby civilians are not merely the objects of rule but are themselves part of the opposition government. Numerous rebel movements, from the Maoist insurgency in Nepal, the Ugandan National Resistance Army, and the Eritrean People's Liberation Front to Hezbollah in Lebanon, the Revolutionary Front for the Liberation of Mozambique, and the Liberation Tigers of Tamil Eelam in Sri Lanka have included civilian actors and organizations in their governing apparatus (Huang Reference Huang2016; Kasfir Reference Kasfir, Arjona, Kasfir and Mampilly2015; Stewart Reference Stewart2018). We discuss the generalizability of our study further in the conclusion.
The Syrian Context
Before turning to our empirical analysis, we first describe the context of the Syrian civil war and the provision of international assistance therein. In doing so, we draw on interviews that we conducted in Syria and along the Turkey–Syrian border between 2013 and 2019. We refer to each interview by number since we offered our respondents confidentiality as a condition for the interviews.
The Syrian war began in 2011 after a violent insurgency erupted in the wake of a popular uprising against the government of President Bashar al-Assad. In late 2011 and 2012, Syrian rebels – made up of Kurds, Islamists and more moderate FSA factions – seized control of many government-held locations in Aleppo, Idlib, Rural Damascus, Raqqa, Hama and Homs, sparking the establishment of rebel governing institutions in hundreds of communities by 2013.Footnote 16 These civilian-led local councils (LCs) aimed to fill the void left by the regime in opposition-controlled areas by delivering basic relief and restoring public services. They operated alongside local and national military opposition institutions, namely the FSA and its Revolutionary Command Council, a collection of locally organized, loosely associated militias. LCs thus coexisted with their armed counterparts as part of the larger Syrian opposition to the Assad regime.
As the conflict went on, the Syrian opposition became known for its ever-shifting internal alliances and rivalries, which often culminated ‘in territorial skirmishes, assassinations and out-right inter-rebel civil war. Rather than coalescing into a unified rebel front, rebel groups continue[d] to compete for power amongst themselves’ (Pedersen and Walther Reference Pedersen and Walther2018, 6). During the period in which most of the data for this study was collected (2014–2016), rebel groups of varying ideological stripes clashed violently in many communities, wrestling one another for dominance of the areas in question. In some cases, they established service delivery bodies that worked as alternatives to, or competed with, LCs.Footnote 17
In the midst of this fractured military environment, Western donors – and neighboring countries – saw LCs as key civilian stewards of the moderate opposition and supported them by pledging millions of dollars of humanitarian aid. USAID/OTI was at the forefront of international efforts to generate support for the Syrian opposition through foreign non-military aid. The office's mission is to stabilize conflict-ridded countries and to encourage long-term development ‘by promoting reconciliation, jumpstarting local economies, supporting nascent independent media, and fostering peace and democracy through innovative programming’.Footnote 18 USAID/OTI began working in Syria in March 2013, providing US$17.58 million in aid to Syrian opposition institutions by July 2016. This aid was primarily in-kind assistance provided directly to LCs – goods, as opposed to cash – to supply the councils with materials and equipment that helped them deliver a wide range of services to their constituents, including education, road repair, rubble removal, emergency response, public outreach, humanitarian assistance, media service, civil society organizations, water and administrative services (for example, civil records and land titling).Footnote 19
USAID/OTI assistance was expressly intended to bolster public support for rebel institutions through service provision. While the agency was not the only donor active in opposition areas, it was clear from our interviews that USAID/OTI was often the first one to enter a community, and that its distribution of aid generally reflected broader trends in Western donor assistance to rebel-held Syria. As one Syrian interlocutor explained, Western donors understood LCs and their work in service delivery as ‘legitimizing the political opposition’.Footnote 20 According to Brown (Reference Brown2018, 6), the donor logic presumed that ‘the popular support they earned from successful service delivery, in turn, would encourage the public to regard the councils as a legitimate, capable alternative to the Assad regime….These assumptions mirrored dominant practice in liberal state-building models and ‘hearts and minds’ counterinsurgency theories, where effective service delivery is seen as key to securing popular support and legitimacy’.
However, our fieldwork in Syria suggests that intra-rebel contestation made this support much more difficult. Multiple LC members in Northwest Syria cited contestation between armed groups as one of the most significant factors interfering with their work. Violent clashes and attacks by rebel factions forced councils to suspend or postpone the distribution of goods, and the resulting instability prevented council members from reaching some of their constituents. Similarly, insecurity stopped people from traveling to LC offices to obtain assistance, provide feedback and engage with council members, which cut civilians off from local authorities.Footnote 21 As a council leader in Aleppo Province explained, ‘when we work in areas that are more stable, we face fewer problems than in other areas and are able to better meet the needs of the people’.Footnote 22
Syrian Survey Data
To test the effects of aid to LCs in Syria, we use new data on Syrian residents' perceptions of opposition governing institutions. We assembled the data using surveys that were administered by an independent research group contracted by USAID/OTI. The data was gathered in five collection periods: May 2014, September 2014, January 2015, August 2015 and March 2016. A total of 13,657 surveys were administered in twenty-seven Syrian communities (a community is the geographic unit used by OTI and is the level at which an LC operates). These locations generally correspond to cities and towns, although they include a few rural areas (for example, Al-Houla, which refers to a collection of small towns in a rural area). This information was coupled with in-depth interviews of roughly 1,100 people from the communities in the sample. As a result, we have unique and detailed information about residents' lives and views during an active conflict that has been largely inaccessible to scholarly researchers.Footnote 23 This represents a contribution to the literature, since much previous research on the Syrian conflict has been shaped by surveys and interviews with Syrian refugees who have fled to neighboring countries including Lebanon (Corstange and York Reference Corstange and York2018) and Turkey (Fabbe, Hazlett and Sınmazdemir Reference Fabbe, Hazlett and Sınmazdemir2019). Our data feature the opinions of individuals living inside Syria during the war, which offers unique insights into a range of lived experiences at the apex of the conflict. This sample covers opposition-controlled areas; OTI did not provide aid to communities in which the Syrian government remained in control. Details about the sampling process are provided in the Appendix.
We are interested in testing the effect of aid and intra-rebel contestation on public support for local governing bodies. The data contain eight measures of such support, each of which is measured on a scale from 1 (lowest level of support) to 5 (highest level of support). These measure the extent to which respondents: (1) support the LC, (2) believe it should play a role in governing Syria if Assad leaves power, (3) how well the LC communicates its activities to the people and (4) supports their needs, (5) is the best option, (6) is (not) corrupt, (7) listens to people who visit its offices or contact it and (8) prioritizes peoples' needs.Footnote 24 We average these components to create a composite variable, LC Support, and rescale it from 0–1 for ease of interpretation.Footnote 25
Quantitative Analysis
Our analysis features two key independent variables. The first, Contestation, is a dichotomous measure indicating whether there was contestation between different rebel opposition groups during each data collection period within a given community. The relevant opposition groups include ISIS, Jabat al-Nusra, Ahrar al-Sham, Jaish al-Islam, the FSA and affiliates, the Kurdish YPG, and several smaller or more localized factions. We categorize an area as contested when no armed group has unchallenged control. Because there is no standardized record of violent skirmishes in Syria during our study period, we constructed our own measure based on whether there were reports of one or more violent clashes between different opposition groups in each community during each data collection period.Footnote 26 This variable was coded using OTI's community profile reports, news sources, and consultations with aid workers and experts on the conflict. We used three coders; if two disagreed, additional research was conducted to verify a given coding decision. We focus on intra-rebel contestation in our baseline specification because this was the main source of contestation in these rebel-held communities.Footnote 27 However, we also employ an alternative specification in which we use rebel–government contestation in the robustness checks, and control for such contestation in other specifications, as described below.
Our second main independent variable, Aid, is constructed from USAID/OTI's database, which shows that 310 projects were completed during the study period. Of these, 192 projects benefitted the communities in our sample, while the others went to communities that were not surveyed. Twenty-eight projects were implemented at the national level, so we dropped them from our sample for a total of 164 projects. While the vast majority were implemented at the community level with the LC as the recipient, twenty-six activities were carried out province-wide through a provincial council. For those twenty-six projects, we divided the total amount of aid by the total number of communities in the province to obtain each community's share. As noted above, aid mostly took the form of equipment and supplies – water tanks, garbage trucks, school supplies, food and non-food items – that were branded with the logos of the LCs (not USAID/OTI) and were used by the councils to provide basic services to their communities. Projects were carried out simultaneously and varied in duration from 1 month to over a year, though since most involved supplying equipment, the duration is not particularly relevant.Footnote 28 This aid was implemented consistently across communities, though OTI often tailored the type of project to the local community's needs.
Table 1. OLS regression results
Note: robust standard errors clustered by community in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.010
Since this aid could be used over multiple time periods, we argue that it should have compound effects, and continue to influence public support in subsequent data collection periods.Footnote 29 Thus, this variable equals the value of aggregate cumulative foreign aid from USAID/OTI, in US dollars, that a community received from the beginning of the sample until the current data collection period.Footnote 30 Within a given data collection period and community, aid ranges from $0 to $1,712,283, as shown in Table 2 (where it is divided by $100,000.) The distribution of this aid is also displayed in Figure 2, which suggests considerable variation in the location and dollar value amount given during this period. Cumulative aid to a particular community ranges from $0 to $5,382,857, which is summarized in Table 7.Footnote 31 Many Syrian opposition councils lacked alternative means to generate revenue, which, combined with the extreme needs of the war-affected population, suggests that this aid represents a large amount of assistance in this context. Indeed, as one LC member of Aleppo City explained, ‘Any assistance is important…it is a challenge to provide services and help our people without donors. We have no one else to turn to.’Footnote 32
In some models we control for the respondents' basic characteristics, and those of their communities. First, we include dichotomous indicators of the dominant armed actor in a community, which we coded using news sources, OTI's community profiles, and consultations with researchers, journalists, humanitarian practitioners and other experts on the conflict. This variable varies over time within communities. We again relied on three coders to ensure inter-coder reliability. These variables include ISIS, Jabhat Al Nusra or Ahrar al Sham, the Free Syrian Army and the Kurdish People's Protection Units. We also include indicators of whether a respondent is Female, Employed and a Muslim Sunni Kurd.Footnote 33 We then added the respondent's Age, Income Footnote 34 and level of Education.Footnote 35 In additional robustness checks shown in the Appendix, we also add two other groups of variables that capture: (1) overall levels of violence: whether a respondent Feels Safe, and the presence of Coalition Bombing and (2) the LC's selection process: whether it held open Elections, Indirect Elections, officials were Appointed, officials were selected based on ClanTies or whether it was a Sharia Court.Footnote 36
In our baseline model specification, we use ordinary least squares (OLS) regression with community and data collection period fixed effects, as fixed effects can tackle time-invariant heterogeneity across units and thus reduce omitted variable concerns. We first present a parsimonious model with no additional covariates, before adding the controls described above. Robust standard errors are clustered at the community level.Footnote 37 We display summary statistics in Table 2, and show our results in Table 5.
The addition of aid has a modest but positive effect on support for LCs in uncontested areas, as the coefficient in our baseline specification is 0.015 and is statistically significant, as shown in Table 1. Substantively, going from no aid to US$500,000 – the mean amount of aid in our sample – boosts support for rebel governance by 0.075 on a scale from 0–1. The average level of support in our sample is 0.49; thus, we find that aid can produce limited but real change in perceptions of governance.Footnote 38 However, in contested areas, the effect of aid becomes insignificant, as shown in Figure 1, in line with our predictions. We also note that the baseline level of support for LCs is large, positive and significant, indicating that, as we have argued, civilians tend to identify with and support these bodies.
Figure 1. Marginal effects, with 95% confidence intervals
Selection Effects
One potential methodological concern is that aid was not distributed randomly, which could bias our results if, for example, OTI targeted its assistance to the best-performing councils. However, while the larger goal of the aid was to win hearts and minds, accounts of US assistance in Syria suggest the absence of a consistent logic behind the communities targeted by OTI in its distribution of aid (Brown Reference Brown2018, 2–3). OTI followed a general strategy to target aid to areas no longer under regime control with the aim of propping up nascent rebel governing institutions. In keeping with the logic of ‘winning hearts and minds’, aid was deployed with the goal of improving public opinion toward these institutions and away from the regime and, later, Islamic extremist groups. Beyond that broad goal, however, there was no consistent targeting process, except to avoid supporting so-called extremist groups. Rather, the approach was decentralized, disjointed and bottom-up. Community selection was largely driven by the idiosyncratic preferences and connections of individual program development officers. While aid was not given randomly, its distribution lacked a coherent, top-down procedure to systematically select particular communities or councils based on their capacities or political contexts.Footnote 39 We describe this process further in Appendix Section H.
Empirically, while we note that our community fixed effects should help address concerns that certain kinds of locations obtained more aid than others, we also conduct a Kolmogorov-Smirnov test to determine whether aid was distributed differently in contested and uncontested areas. We find that less aid was provided to uncontested areas, implying that a potential selection bias may run against our theory. While these findings do not eliminate potential selection effects given the observational nature of the data, they help to mitigate them.
Finally, we use an instrumental variables approach. We exploit the fact that a large portion of the aid we consider was delivered across the Turkish border into Syria, from where ‘it [went] where it [could] be most easily provided, not necessarily where it [was] most needed’ (Gill Reference Gill2016).Footnote 40 Thus we expect that communities located closer to the Turkish border received more aid, due to their proximity to the source, whereas those further from the border received less aid. We therefore use distance to the border as an instrument for aid flows. We measure distance in three ways. First, we used the shortest distance from a community to the Turkish border. Secondly, we used the distance through Lebanon for relevant areas. Thirdly, we used the distance to the closest crossing station. We include the same control variables as in our main specifications and robustness checks, along with survey period fixed effects. We find that we do not have a weak instruments problem, as the Donald-Cragg statistic exceeds the critical value.Footnote 41 Column 3 in Table 5 shows that our main results are robust to this specification.
For distance to be a valid instrument, the exclusion restriction must be satisfied; that is, distance to the border should only affect support for the LCs through its effect on aid flows. While we cannot prove that the exclusion restriction holds, we argue that this is a plausible assumption, though we acknowledge that violations of the exclusion restriction are possible. Of particular concern is that distance could affect LC support through its impact on migration, as those closest to the border may be more likely to migrate, which could in turn influence attitudes. We thus also present the results of an instrumental variable analysis that limits the period of study to the last two survey waves, given that these occurred after the border was closed to migrants in November 2015.Footnote 42 Our results hold under this specification, as shown in Appendix Table 3.
Alternative measure of Contestation
Another potential concern with our analysis might be that contestation is not randomly assigned. Yet rebel groups often target areas due to well-known historical, geographical and religious factors, many of which are absorbed by our fixed effects.Footnote 43 We also add control variables that capture several potential confounding variables, including how LC leaders were chosen and safety concerns.Footnote 44 This approach gives us confidence that our respondents’ basic characteristics do not significantly differ in contested versus uncontested areas.Footnote 45
As an additional robustness check, we also employ an alternative measure of contestation. While we have focused our analysis on contestation between rebel groups, our theory should also apply to contestation between the national government and rebels. In particular, contestation along the war's master cleavage – between the Syrian regime and rebel forces – should condition aid's effect on opinions of the national-level opposition body, the Syrian Interim Government (SIG), as well as the LCs. USAID and other donors worked with the SIG, the opposition's executive representative that sat in exile in Turkey, to co-ordinate assistance for LCs and ‘help the opposition leverage this cohesion at high-level political negotiations’ (Brown Reference Brown2018, 7). One American official described how aid was branded with the insignia of the opposition leadership in order to convey to those receiving goods and services that the national-level opposition was ‘actually relevant inside Syria’.Footnote 46 In 2016, the then-interim deputy prime minister stated that the SIG's multi-pronged mission was ‘to provide [citizens] with institutions that could normalize their lives, provide them with basic utilities…also to gather help and assistance for the Syrian people…It was also a way of really trying to defy the regime's legitimacy in the eyes of the international community by providing the alternative or the legitimate alternative’.Footnote 47 Our theory anticipates that aid could improve public opinion of the SIG and LCs, except in areas experiencing national-level contestation.
To test this proposition, we constructed a dichotomous measure of national-level contestation for each community during each data collection period, based on whether pro-government fighters attempted to retake the community from the opposition through ground attacks. If rebel sovereignty was not violently challenged by pro-government forces, we considered the area to be uncontested.Footnote 48 We also constructed a dependent variable similar to the one used in the main analysis, but based instead on survey questions asking about public support for the SIG. Our survey data contain six measures of support for this body, each of which is scaled from 1 (lowest level of support) to 5 (highest level of support). These include measures of the extent to which: (1) people support the SIG and believe that it (2) should play a role in governing Syria if Assad leaves power, (3) communicates its activities to the people, (4) supports the needs of the people, (5) is the best option and (6) is (not) corrupt. We again average these components to create a composite variable, in addition to examining each question individually, and rescale them from 0–1. Note that this variable has fewer components than the local-level measure because USAID/OTI asked fewer questions about the SIG. The results, shown in Column 4 of Table 5, corroborate our main results by demonstrating that aid only improved support for both LCs and the SIG in communities uncontested by pro-regime forces. While our alternative measure of contestation is also not randomly assigned, it is reassuring that we find similar results using both measures.Footnote 49
Qualitative Analysis
Our quantitative findings suggest that foreign aid amplifies a civilian opposition institution's governing reputation in uncontested spaces. We explore the mechanisms underlying our theory qualitatively at the local level by conducting comparative case studies in two opposition-controlled communities: Aleppo City in Aleppo Province and Saraqeb in Idlib Province. While both LCs were major recipients of US aid, and arose in the wake of the popular uprising against the Assad regime, they differed based on the coercive context in which they operated. Aleppo City featured high levels of contestation, while Saraqeb remained militarily uncontested during our period of study, enabling its council to translate aid into popular support, whereas its counterpart in Aleppo City could not.Footnote 50
Moreover, we show through detailed interview responses that all three of our mechanisms contributed to this outcome. First, in Aleppo, people frequently noted the need for more military protection and equipment, expressing the view that the LC suffered from an inability to protect itself and those it governed. Secondly, interviewees noted that armed Islamist groups had an advantage over the LC when it came to service provision on account of their capacity to protect the goods they had on offer and to out-compete the LC in servicing some neighborhoods and sectors. Finally, due to the variety of service providers in the city, it was often difficult for residents to ascertain who had delivered particular services. In contrast, in Saraqeb, despite episodic disgruntlement with the LC's politics, people recognized the council for the clear and consistent aid it provided and noted that it was protected by the Saraqeb brigades.
Aleppo City
We elected to study the Aleppo City Council (ACC) for several reasons. First, during our study period, USAID/OTI channeled more than $5.6 million into Aleppo City in the form of equipment, stipends and trainings across twelve sectors, making it the largest US aid beneficiary in Syria at the time. One US official stated that the ACC was ‘a high capacity, high performing council that could take money and translate it into tangible activities on the ground’.Footnote 51 The ACC thus represented an ideal test case for the translation of aid-enabled service delivery into support for the institution. Moreover, the council emerged as a popular grassroots institution, an unmistakable product of the city's revolutionary politics. Our interviewees, native to neighborhoods that aligned their politics with those of the rebellion, recognized in the ACC an organization embedded in local social networks and attuned to the community's revolutionary aspirations. This recognition furthered its potential to become an authoritative governing body. Aleppo City thus represents a hard case for our theory, since people were arguably predisposed to consider this LC favorably.
However, the ACC sought to govern in a highly contested space, in which no armed group could claim a monopoly over the Aleppine effort to remake Syrian politics through revolution. At the height of the Syrian civil war, Aleppo's cityscape served as a microcosm for the larger conflict.Footnote 52 An inventory of the armed groups that occupied space within the opposition-held eastern part of Aleppo City between 2014 and 2016 includes every major faction of significance across the country, while the Assad regime's forces controlled the remainder of the provincial capital.Footnote 53 Several Syrian interlocutors described their surroundings as being highly contested. In 2014, one resident explained the interaction between the city's multiple, competing armed groups: ‘If they are confronted in the street, they will fight. This is the law of the fittest, and the fittest one is the armed one’ (2014_2_ALE 001). Another concluded, ‘we now live under the rule of weapons, which makes accountability impossible’ (2015_1_ALE 002). A third described the insecurity and unpredictability that came from this dispersed coercion: ‘There are a lot of militia in the area. People don't feel secure. If someone has a problem with you, he might get the help of any armed group to kidnap or kill you’ (2015_1_ALE 007).
This unrelenting armed competition created an environment in which the ACC, despite the aid-enabled services it provided, proved unable to deliver the most critical public good of all: security. Residents frequently noted the importance of military protection and security, and unfavorably compared the non-military aid the ACC received to ‘harder’ forms of assistance it often required. For example, one interviewee stated, ‘We asked…for equipment to pull people from the rubble, but [the donors] said ‘no’ because it is military equipment. What are we supposed to give those working in rescue…Roses?’Footnote 54 Indeed, interviews in both 2014 and 2015 indicated that the ACC was in a precarious position amid the various armed groups operating in the area. While the aid the ACC distributed was helpful, respondents indicated that what it really needed to govern was ‘a protection force’ (2014_3_ALE 003) because ‘arms always have the strongest voice’ (2015_1_ALE 007). The ACC proved incapable of reliably leveraging coercive power on its own behalf, let alone on behalf of ordinary civilians, which undermined its ability to portray itself as an authoritative governing body. Respondents commented on the resulting impotence: ‘the opposition brigades are the ones who hold the local council's members accountable and might threaten them if they are not satisfied’ (2014_2_ALE 005).
Moreover, many interviewees noted that the ACC was cut out of service provision in various areas due to the stiff competition posed by other rebel organizations, some of which jointly maintained armed and civilian administration bodies. For example, one interviewee noted:
The LC cannot provide bread properly in the neighborhood because it is controlled by the General Services Administration (affiliate of Jabhat Al-Nusra), and [it is] not active in the health sector either. These are the reasons that made me come to the conclusion that people are not satisfied with the LC. Furthermore, the LC cannot interfere in water services because these services are also controlled and provided by the GSA, and the electricity has been cut for three months from the neighborhood (2016_3 ALE).
The interviewee further noted that both the lack of security provided by the ACC and the services delivered by other groups together shaped the perceptions of the council in the community: ‘what I'm seeing is that the GSA is trying to control the service sector completely and is marginalizing the ACC from doing its job with the excuse that it is being supported by external sources – “infidels”’. The interviewee elaborated:
What controls our neighborhoods are coordination and administrative bodies which are Jabhat Al-Nusra affiliates, or follow JAN's ideology…and they are taking the role of the service authorities…Anything that does not abide to their way of management is banned and must not exist, like when the ACC opened the schools, they forced the schools to segregate the students into male and female beginning of the 2014 academic year. Therefore, you find many simple citizens afraid to communicate with the ACC to avoid any reaction from those Islamists. The ACC cannot work individually in our area, and it doesn't communicate with people at all, because the Islamists are controlling the neighborhood and the military power the LC lacks the Islamists have (2016_3 ALE).
Another respondent, discussing the ACC's limitations, also contrasted it with Jabhat al-Nusra and described the degree to which the latter's coercive capacity enabled it to capture the management of public goods like water and electricity, some of which had been previously supported by foreign aid: ‘In the early days, there were lots of thieves and chaos and even the infrastructure was robbed. As the local council lacks military force, it failed to maintain security. So Jabhat al-Nusra took control of the water institution and expelled the thieves there…The same applies for electricity’ (2015_1_ALE 007). Another interviewee similarly pointed to differences between the ACC and Nusra's GSA, which operated services in an adjacent neighborhood:
[T]he services provided there are great, the electricity situation is always better than ours, so is the water, and when there are water cuts, they send tanks to fill water for the residents. Additionally, the GSA dug up wells in the streets of that neighborhood, to provide for the residents, and it provides electricity to the water pumps that are connected to those wells. I think it is very positive to see that a military power also has executive powers, it does what it wants and is not afraid of anyone and does not wait for anyone's approval from abroad, which is not the case with the ACC (2016_5 ALE).
The Aleppo City Council's reliance on external assistance may have undercut its reputation vis-a-vis competing service providers, whose independence and autonomy sharply contrasted with a civilian institution that was dependent on foreign support.Footnote 55
Even when the ACC did provide quality services, the violently contested environment in which it operated created opportunities for other groups to take credit for its aid-funded work. According to a USAID/OTI report, the ACC's competitors, such as Jabhat al-Nusra's services administration, ‘undermine, and even take credit for, ACC projects’.Footnote 56 One interviewee confirmed, ‘often the GSA [Nusra's services administration body], through the mosque clerics, takes credit for the ACC's work, twist[s] facts, and say[s] that the GSA…did the work done by the ACC’ (OTI 2016, 28).
Ultimately, the ACC could not complement its political programs and service delivery with credible coercive capacity. Instead, it remained only one of several contenders for governing power within the city, unable to protect itself, its assets or its constituents. As one of the respondents quoted above described, ‘the ACC in our neighborhood is a weak LC and cannot stand up to the General Services Administration, so how can I believe in the security of my area and country when the ACC that is supposed to take care of us, is afraid of the military [groups] or those that are supported by them?’ (2016_3 ALE). Thus the ACC's well-supported capacity to deliver services could not compensate for its precarious position amid acute inter-rebel contestation; as a result, it failed to translate aid-funded service provision into a reputation for authoritative governance.
Saraqeb
A different story unfolded between 2014 and 2016 in Saraqeb, a revolutionary bastion in Idlib Province. Like the ACC in Aleppo City, the Saraqeb LC was one of the key targets of American support, receiving nearly $2.5 million during the study period from USAID/OTI for emergency response, humanitarian aid and waste management. Like in Aleppo, the council grew organically out of the social solidarities underpinning the town's grassroots struggle against the Assad regime. This case, however, featured non-contestation between armed actors. Thus the influx of American aid amplified the council's capacity to serve its constituents.
The town's dominant armed groups – the Islamist group Ahrar al Sham and local FSA affiliates – did not violently contest one another. Instead, they exercised substantial influence over the membership and decision making of the civilian institution in their midst. These growing interactions between armed groups and council members could be read as the very kind of symbiosis that lends itself to generate order out of anarchy. It is, as North et al. (Reference North, Wallis and Weingast2009) explains, precisely the exclusionary aspect of resource sharing that incentivizes powerful actors to co-operate rather than compete. Even Saraqeb residents who were critical of military involvement in politics betrayed a clear recognition that the armed groups on the ground translated their peaceful coexistence with one another into the kind of protection that is critical during wartime, which was facilitated by aid (SAR_2014_2_002).
The Saraqeb LC, in the shadow of its armed patrons, moved considerable capital in the service of much-needed public goods, which often came from USAID and other donors. Interviewees described the Saraqeb brigades as focused on the fight against Assad and securing their town, which included protecting the LC's service provision:
These fighters are the sons of our area. [They] will work to protect people from murders that have spread in the town and the surrounding region recently. In general, there is coordination and cooperation between the Front of Saraqeb Revolutionaries and other brigades with the Local Council (SAR 2015 2 009).
In the absence of contestation, it was clear who was providing goods and services. ‘We see their [the LC's] activities here’, said one interview respondent (SAR_2014_2_002), while another reported: ‘Yes, there's responsiveness, and a lot of it – they support the entire community’ (SAR_2014_2_003). A third respondent concurred, pointing to the council's efforts to improve residents' access to electricity after the need was brought to its attention (SAR_2014_2_004). A fourth respondent echoed this point: ‘They say, we will offer this project to so-and-so supporting organization, and if they approve it, we will work on it. So they do respond to people and try their best within their capabilities’ (SAR_2014_2_006).
Moreover, by 2016, due to the lack of competition with other armed service providers, the LC reaped the reputational benefits of the aid-funded services it provided. One interviewee stated that his positive view was because, ‘The LC is currently in charge of providing services and taking care of families of martyrs and other services’ (SAR_2016_5). Another stated, ‘A good percentage of the people in the city support the LC because of the services it provides, including communications, water, sanitation. Civilians in general support the LC, which contributes to the improvement of the city through the services and projects it implements. When the LC is active and provides more services, the number of its supporters increases’ (SAR_2016_4). A third respondent explained, ‘What makes for a good local council is the ability to manage the city's services in all sectors, including basic services, education, and regulating prices’ (SAR_2016_3). Some interviewees specifically stated that they believed that this was the proper role of the LC, rather than armed groups. One exclaimed: ‘Accessing services for civilians through the LC is much better than accessing them through military groups’ SAR_2016_2).
The LC's success in disbursing aid enticed armed actors to seek influence over its membership and work. As a result, popular protests erupted during the study period, but by mid-2015, residents described modes of coexistence, even cooperation, between military and civilian actors in Saraqeb. The regime continued to batter the town with aerial bombardment but, at least on the ground, the absence of violent contestation and militant actors’ inclination to support their civilian counterparts created a space in which the work of civilian governance – including much-needed service delivery – could unfold under ordered conditions. In this sense, foreign aid proved catalytic in consolidating a council whose existence, membership and activities served the militant project and thus had the space and capacity to establish itself as an authoritative provider of public goods.
Alternative Explanation
A potential alternative explanation is that perhaps relative weakness rather than contestation can explain our results. Perhaps the LC in Aleppo needed support in hard power rather than economic aid given that it was facing the powerful al Nusra. However, we argue that the LC's relative weakness cannot explain our findings because contestation creates a number of problems that cannot be solved via institutional strength. Our first mechanism involves the importance of hard power in the midst of contestation. Even if an LC is materially strong, hard power is still likely to be in greater demand than non-lethal aid in contested areas as entities vie for control and create insecurity for those living in their midst. In an uncontested area, even a weak council can offer protection because the threats to its constituents are limited. Our second mechanism involves the vulnerability of aid in the face of contestation. Absent contestation, aid would be less vulnerable to sabotage, theft and assault. This remains true regardless of the LC's strength. Our third mechanism involves attribution problems. If there was no contestation, attribution of aid would be relatively clear regardless of the LC's strength, since only one group would be in the business of governing.
In Aleppo, Nusra could not claim the ability to create a secure environment or to address all attribution errors. Any military superiority it possessed may have enabled it to guard its work from destruction more effectively but, in a highly contested environment, it remained vulnerable to asymmetric attacks. Indeed, accounts indicate that Nusra had trouble providing services effectively in this environment: ‘A simmering proxy war over service provision emerged that would endure until the end of rebel control in Aleppo….[Nusra] would answer that it was at least trying [to provide services], but was severely constrained by circumstances in the city’ (Donker Reference Donker2020).
We also attempt to account for this variable empirically. We now proxy for an LC's strength using the number of months it was operational, based on the assumption that LCs that existed for a longer period of time did so both due to their initial capacity and the opportunity that time afforded them to gain additional capacity. Our results hold while controlling for this variable, as shown in Table 6.
Conclusion
We have demonstrated that foreign aid is more effective at ‘winning hearts and minds’ during civil war in uncontested communities. Aid does not generate support for governing institutions in contested communities because insecurity and competition from other actors undermine the ability of these institutions to offer protection to those under their writ, to maintain efficacious and sustainable services, and to receive credit for their work. Using novel data on public perceptions, aid flows and armed contestation during the Syrian civil war, we find qualitative and quantitative evidence to support our claims. Aid distributed by USAID/OTI improved support for local governing councils only in areas lacking intra-opposition contestation.
The importance of control over the use of force – which is well established in the literatures on civil war and state formation – has significant implications for foreign efforts to promote domestic governance. Our findings suggest that aid may improve the credibility of rebel governing institutions in the absence of contestation, but it may also facilitate the consolidation of politics at odds with the democratic ideals that many Western donors hold dear. Civil institutions supported by their armed counterparts may become more credible governors. At the same time, they may also adopt the kinds of predatory, exclusionary practices associated with limited access orders, thereby undermining other aspects of ‘good’ governance. These forms of institutional capture have been identified in other contexts and suggest the need for further inquiry into the relationships between aid, civil–military relations and governance outcomes (De Waal Reference De Waal2009; Mukhopadhyay Reference Mukhopadhyay2014; Reno Reference Reno1997).
While we tested our theory in the context of rebel governance in Syria, we expect our findings to travel to other cases as well. Foreign governments and international donors have funneled substantial non-lethal assistance to both state and non-state actors in conflict-affected countries. We anticipate that foreign aid injected into highly contested settings will prove ineffective at boosting the popular reputation of governing institutions regardless of whether it is delivered to state governments or insurgents, as the mechanisms we have outlined are unique to neither Syria nor to rebel groups. Weak states – at the national and local levels – face similar difficulties in areas marked by banditry, insurgency and terrorism, all of which undermine their capacity to manage violence. Aid designed to support these states as part of campaigns to subdue rebellion and violent extremism are likely to fail for the same reasons we have laid out with respect to insurgent rule in Syria. States and counter-states face comparable challenges in winning popular allegiance through aid when politics are riven by violence; in fact, states may face additional hurdles when delivering aid in uncontested areas, as described above.
However, our findings also highlight promising directions for future research, as our results are not likely to hold in all settings even as they may be more robust in others. For example, the type and degree of contestation may condition our findings. The Syrian context features especially fragmented territorial control with a large number of competing armed challengers, which likely makes the Syrian case a hard one for our hypothesis. Aid-enabled services may be easier to attribute to a given actor with fewer actors in the space, in which case aid would be more likely to positively affect political attitudes.
The Syrian civil war is also exceptionally violent, which may make our mechanisms especially salient. In less violent spaces, people may be less concerned about their security relative to other public goods and services, and aid-enabled goods and services are less vulnerable to sabotage or destruction. It is also possible that the effects could vary depending on the type of aid and the identity and goals of the donor. For example, aid given as cash rather than equipment could be stolen or diverted more easily, even as it might be welcomed for its fungibility.
Further, US aid was managed remotely, from over the border in Turkey, on account of the high levels of violence that made the security situation impermissive to a sustained Western donor presence on the ground. The remote management of aid made this intervention less coherent than it might otherwise have been (Howe Reference Howe2016). So, too, did the ever-shifting concerns of the US government with respect to the Syria conflict and its intervention therein, which evolved from a focus on removing the Assad government to a subsequent concern with countering violent extremism. The absence of a more concerted strategy, while not exceptional to US interventions abroad, was particularly striking in the Syrian case,Footnote 57 making it, again, a hard case for our hypothesis. If donors were to adopt a more concerted approach to delivering aid, we would expect the positive results of aid to be stronger in uncontested areas. However, a more coherent and consistent donor approach would be unlikely to overcome the problems that stem from contestation that we have highlighted.
Another interesting avenue for future work would be to explore additional efforts to ‘win the hearts and minds’ of the surrounding population. Both military and civilian entities have roles to play in this endeavor: military forces provide security, and are more likely to win support if they do not abuse the population, while civilian actors provide other goods and services. While we focus on the civilian side, further research might explore the efficacy of civilian vs. military wings of a rebel movement and gauge the relative importance of each in winning hearts and minds. A related fascinating question would be to explore the degree to which success in winning hearts and minds on the civilian side of a rebel group advances its military cause.
This study also has important policy implications. Foreign interveners continue to channel non-military aid into violently contested settings hoping that it will ‘win hearts and minds’. Our results suggest that territorial control should be considered a prerequisite of international aid in civil war. The introduction of non-military assistance into contested areas may actually prolong conflict by keeping individuals, institutions and hope afloat without injecting the kind of hard power required to create credible rule (Kuperman Reference Kuperman2008; Lynch Reference Lynch2016). Finally, this research speaks to the tendency of Western donors to put their faith in the power of ‘the local’ – communities, institutions, and organizations – to transform conflict and build peace.Footnote 58 Yet the micro-politics of rebellion involve a number of cleavages that can complicate and undermine even the most locally driven intervention (Kalyvas Reference Kalyvas2003). We illuminate the need for careful consideration of the various forms of local contestation and cooperation at play, both between rebel groups and between military and civilian elements within a given opposition movement as part of any meaningful analysis about the impact of foreign aid in conflict settings.
Supplementary material
Online appendices are available at https://doi.org/10.1017/S0007123421000156.
Data availability statement
Replication data files are available in Harvard Dataverse at: https://doi.org/10.7910/DVN/QOLOQ6.
Acknowledgements
We thank the participants of seminars at Columbia University, New York University, Tufts University, and the ISA and IPES Conferences, as well as Sarah Bush, Rana Khoury, and Noah Zucker for helpful comments and conversations. We gratefully acknowledge funding and data from OTI. We received excellent research assistance from Max Marder and Jasper Yang. All remaining errors are our own.
Conflicts of interest
Allison Carnegie, Kimberly Howe, and Dipali Mukhopadhyay have consulted for OTI in the past.
Ethical standards
This research was exempt from Columbia and Tufts' ethical approval procedures because it relies on third party data.
