State-Building,Collective Efficacy,and the Co-Production of Public Goods in Rural Africa

Introduction

Understanding why some communities are able to resolve local challenges – be it repairing a school roof, organizing monthly village cleanups, or building a health clinic, as described above – is critical to understanding rural development, and the question holds both academic and practical implications. In the context of the developing world, where many public goods are co-produced between citizens and the state, local capacity for collective action can determine whether a community has access to clean water, primary education, or basic healthcare (for example, Hern, 2019). These questions have been amplified in recent decades, as international donors have jumped on the idea of participation and social capital as vehicles to enhance community ‘ownership’ of development (Grootaert & van Bastelaer, 2002; Wiemer, 2021).

We join this work by drawing attention to the concept of collective efficacy: the shared perception that a group or community can coordinate their action in pursuit of a desired outcome. Drawing on work in sociology, which has largely studied collective efficacy in the context of urban America (e.g., Sampson, 2003), we argue that collective efficacy also strongly predicts the capacity for local-level collective action and development outcomes in rural communities in the Global South. Critically, collective efficacy provides us with an assessable mechanism that can translate otherwise neutral ‘stocks’ of group-level properties, like social capital, into action.

Why do some rural societies have higher levels of collective efficacy than others? We propose that variation in levels of community collective efficacy can be explained by distinct histories of state-building. More specifically, we argue that over time, interventionist states that orient citizen action towards the center undermine local collective action. In rural areas peripheral to the state, where collective efficacy is most likely to matter for development, some of the most prominent theories of state-building (Boone, 2003, p. 37) treat the periphery as homogenously ignored by the state. However, more recent work has shown that even small state interventions can have outsized effects in the hinterlands (Nathan, 2023), and an older literature on African state-building proposed that colonialism introduced a key source of variation in levels of state interventionism (Crowder, 1968; Miles, 1994). Bringing these theories together, we argue that even in rural hinterlands, the degree of state intervention can vary in systematic ways depending on colonial legacies, ultimately impacting contemporary local levels of collective efficacy.

In short, we argue that communities that inherited a statist orientation will see lower levels of collective efficacy in rural peripheries today as a result of their emphasis on more centralized and interventionist methods of governing. We argue that this is the most common outcome for communities that were governed under colonial policies of direct rule, generally associated with the French. In contrast, where the state has historically been less interventionist - typically those administered by colonial policies of indirect rule that are broadly, but not exclusively, associated with the British - rural citizens in the periphery should be relatively more empowered to think and act locally, which, over time legitimizes and reinforces local capacity to coordinate collective action.

We investigate these legacies by looking at fifty paired villages along the Ghana-Togo border, areas considered to be hinterlands to the state. We show that collective efficacy is higher in Ghanaian villages, where the colonial and post-colonial state has been consistently less interventionist, than in their counterparts on the Togolese side of the border, where the last century of state-building followed a largely statist model. Building on a mixed-method research design that pairs an original survey, archival material, and long-form interviews with village chiefs and elders, we show that Ghanaians  are more likely to believe that their fellow villagers will take action to correct community problems—everything from disciplining an errant child to keeping the village tidy. ³ We also find that Ghanaian villages are more likely to report having actually resolved a recent local problem. Our results hold under numerous robustness tests, and we demonstrate the argument’s generalizability beyond the Ghana-Togo border by exploiting variation in the degree of state intervention elsewhere in the region as well as sub-nationally within Ghana.

These findings build on a recent interest in behavioral historical legacies (Acharya et al., 2018; Charnysh, 2019; Pop-Elesches & Tucker, 2017) as well as work examining how variation in historical forms of governance shapes enduring preferences over social norms (Lowes et al., 2017). We contribute to the study of collective efficacy more broadly by demonstrating its relationship to significant development outcomes in the Global South and examining its origins in institutional arrangements—something that has not been seriously investigated by the extant literature. Finally, this research addresses the call by Boix and Posner (1998) (amongst others) to more robustly theorize the mechanisms linking social capital to government performance.

Collective Efficacy

Our core expectation is that different repertoires of state-building have left a lasting impact on collective efficacy in the hinterlands of sub-Saharan Africa, which in turn has affected rural capacity for collective action and the co-production of local public goods. Drawing upon the literature from sociology (Bandura, 1986; Janowitz, 1975; Sampson, 2003; Zaccaro et al., 1995), we define collective efficacy as shared expectations that a group or community can coordinate their action in pursuit of a desired outcome. At the heart of the concept of collective efficacy is the existence of shared expectations. The belief that others in one’s community will act in the face of a common dilemma generates reinforcing expectations that become part of the community’s belief structure. This makes the perception of others’ willingness to act a critical collective resource (Zaccaro et al., 1995, pp. 310, 312). In groups marked by high levels of collective efficacy, group members are more likely to contribute to the common good because they believe their actions will be reciprocated, and they will not be taken as a ‘sucker’ (Kerr, 1983). Such groups may even come to define and interpret group outcomes differently. For example, a sports team with high collective efficacy may be more likely to explain a loss through the lens of attributes of the specific match, as opposed to structural causes attributable to the team itself (Zaccaro et al., 1995, p. 310).

Like theories of social capital (Coleman, 1994), collective efficacy emphasizes how trust can facilitate community action. However, as Sampson argues, the existence of collective efficacy is not dependent on cross-cutting or dense social networks, but rather can exist under various network structures (2003, p. S58). Collective efficacy is thus not a rival explanation to theories of social capital, but nor are they conceptually equivalent (see Asari, 2013). Although collective efficacy depends on some level of social capital, it offers a more precise theoretical mechanism by focusing on how and when communities are able to act together. Further, the concept of collective efficacy improves upon the associational focus of much of the social capital literature. Many of these theories assume that all groups in society are equally efficacious (Putnam, 1993, 2000), while those that are more discerning, such as Tsai (2007), oftentimes end up analyzing only a very narrow set of associational groups. Collective efficacy encourages us to move beyond a focus on ‘stocks’ of social capital and to focus instead on collective efficacy as a means to assess how effectively various groups or communities are able to overcome collective action problems.

This approach centers group agency and also highlights how collective efficacy is distinct from individual self-efficacy, or the idea that individuals believe themselves able to undertake a given behavior necessary to obtain a desired outcome (Bandura, 1977, p. 194). Certainly, self-efficacy and collective efficacy are related and likely interact in theoretically important ways. For example, recent experimental interventions have tried to bolster self-efficacy in an effort to improve collective outcomes like education (Lieberman et al., 2014; Lieberman & Zhou, 2022). The null findings produced by these studies may reflect the emphasis on early socialization, as scholars suggest that the most critical source of self-efficacy is the family and, notably, what they see modeled at a young age. Similarly, social aggregations and collectivities can be more or less efficacious depending on how individuals embedded within them perceive their social environments to be capable of collective action (Bandura, 1986). In other words, actual outcomes can be less formative than perceptions of outcomes as an individual develops beliefs about ‘system responsiveness;' the idea that one’s broader environment enables one to produce desired outcomes over their lifespan (Gecas, 1989, 300–301). Of course, an individual growing up in a community that models collective efficacy may socialize individuals to have higher rates of self-efficacy (Bandura, 1977, pp. 195–200), but we caution against the conclusion that collective efficacy is merely aggregated self-efficacy. Collective efficacy is a group-level property that - while reinforced by individual behavior – needs a collective catalyst and is not merely the sum of individual efforts.

Although existing literature has shown that groups with higher levels of collective efficacy are less likely to experience crime (St. Jean, 2007), more likely to have successful outcomes on specific tasks (Tasa et al., 2007), and are less likely to experience negative psychological effects after group-level trauma (Muldoon et al., 2017), there is less work on the origins of collective efficacy. Some research has shown that collective efficacy is built upon successful past performance (Seligman, 1975), though these findings were produced within an experimental context as opposed to within organic communities. Therefore, this paper contributes to the study of collective efficacy in several ways: first, by looking at structural factors that may influence levels of collective efficacy at the local level, second, by bringing the concept to the African context, and third, by linking collective efficacy to development outcomes, something of critical importance to communities across the Global South.

Repertoires of State-Building in West Africa

We propose that one of the key institutional drivers of varying levels of collective efficacy in the Global South is found in different repertoires of state-building. Early work focused on variation in colonial state-building drew a distinction between strategies of indirect versus direct rule (Crowder, 1964). The key difference between these philosophies lies in their relative orientation toward the center. Policies of direct rule—strongly associated with the French colonial state—were built on the foundational idea that formal rules promulgated from the center should supersede local autonomy. Direct rule was thus built upon a clear chain of command (Lange, 2009, p. 28). Accordingly, the French believed that political authority should be firmly vested in French colonial officers, with indigenous chiefs serving as subordinate state agents who followed the directives of colonial officers. As Mazrui (1983) argues, by emphasizing Frenchness as a unifying principle, the French colonial state discouraged local decision-making and capacity-building as colonial subjects were assumed to be on a path to assimilation into French culture (Crowder, 1964, pp. 199–202).

In contrast, indirect rule – associated most strongly with British colonization of Africa – was broadly predicated on the idea that British colonial administration should be constructed within the framework of local tradition (as understood, of course, by the British). The British were not ideologically committed to importing British culture or political systems into their non-settler colonies and were far more interested in identifying, cataloging, and homogenizing indigenous ethnic groups so that they could be organized under ‘legitimate’ indigenous leaders. As a consequence, the British colonial state remained concentrated in colonial capitals, with a ‘miniscule’ administrative presence, meaning that it possessed minimal infrastructural power and failed to robustly regulate society or implement policy (Lange, 2009, p. 33). Consequently, many communities saw minimal British interference in local issues as long as they met British demands for taxation, labor, and cash crop production (Crowder, 1964, pp. 198, 169).

Of course, these are ‘ideal types’ of colonial administration. In practice, as has been documented elsewhere (Gerring et al., 2011; Lawrence, 2017; Müller-Crepon, 2020), colonial administration varied both across and within colonies. In most areas of Africa, the colonial state was thin on the ground, and in some regions (e.g., the hinterlands) the French administration relied heavily on indigenous intermediaries. But even in these areas, the French granted authority in a much less flexible way, appointing intermediaries based on their qualities of “Frenchness” and insisting that the intermediary’s authority was derived from his position within the French colonial administration rather than from his legitimacy vis-à-vis the society he governed (Wilfahrt & Letsa, 2023). In contrast, the British colonial administration spent considerably more time identifying legitimate ‘traditional’ chiefs and bolstering their local authority; something in which the French expressed comparatively little interest (Chiabi, 1997, p. 27). To date, work has linked these differences to variation in contemporary identities (Ali et al., 2018; McNamee, 2019), economic development (Lee & Schultz, 2012), and everyday social order (Asiwaju, 1976; MacLean, 2010; Miles, 1994).

Postcolonial states generally reinforced these governing tendencies, meaning that Francophone states were more likely to subsume powers of local government to the control of the center (Tordoff, 1994, pp. 556–8). As noted by Reyntjens (1988, p. 74), the French Fifth Republic laid out a constitutional template for the strong, centralized state apparatus that persisted in many of its former colonies. Similarly, LeVine (2004, pp. 97–98) identifies persistent norms of exercising power in Francophone Africa that favors structuring power hierarchically toward the center. In contrast, as the British colonial state emphasized local authority, so did—at least nominally—many postcolonial British colonies. While Tordoff (1994, p. 557) notes that in the immediate postcolonial period former Anglophone states tried to centralize political control authority by homogenizing and subsuming the authority of decentralized political institutions, “the results were mostly disappointing.”

Highlighting how colonial legacies persist in the postcolonial period, authors have discovered differences in political behavior in otherwise highly similar border communities. For example, Miles (1994, pp. 280–81) finds that in francophone Niger, development projects were more likely to be initiated and managed by the central state, whereas in neighboring anglophone Nigerian villages, development initiatives were more often locally initiated, with the local state a meaningful site of political coordination and contracting for citizens. MacLean (2010, p. 213) similarly finds along the Ghana-Côte d’Ivoire border that when asked what was most important for village-level decision-making, Ivoirians named ‘big men’ who resided outside of the village while Ghanaians pointed to village-level political institutions.

An alternative literature departs from this focus on colonial legacies to highlight enduring patterns of subnational state-building in Africa that defy colonial heritage. Most importantly, Boone (2003) has argued that variation in the nature of rural society strongly shaped postcolonial state-building strategies across West Africa regardless of colonial heritage. Boone argues that states treated cash-crop producing regions in a fundamentally different manner, adopting a variety of different strategies depending on the nature and economic position of rural society vis-à-vis the state. In contrast, economically peripheral areas were uniformly “non-incorporated,” with Boone describing the regime as “forsaking state building” in these regions altogether (35). More recently, Nathan (2023) contributes to this framework by investigating such “non-incorporated” areas, or “hinterlands,” agreeing with Boone that while states tended to ignore peripheral areas, when it does act in these zones, its interventions have profound impacts on society.

We build upon this framework by re-introducing the importance of legacies of colonial approaches to state-building in the hinterlands. In agreement with Nathan’s (2023) broad point that the state can have the largest impact in the most peripheral regions, we further posit that the nature of these interventions vary systematically by colonial heritage. While colonial legacies may matter less for key economic zones where power is continuously and hotly negotiated between the state and powerful economic actors (Boone, 2003), in the periphery where the state is less involved, the regime is more inclined to continue the ideological strategies developed and implemented under colonial rule. Thus, we expect that in the hinterlands of former French territories, where the state has long been more interventionist, the state is more likely to rule “directly” by undermining local power structures and emphasizing the authority of the central state over local autonomy, legitimacy, and decision-making.

We hypothesize that the cumulative effect of an orientation in state-building toward the center—which shifted the locus of decision-making away from local communities—more extensively undermined the legitimation of local arenas of political contestation and action. Over time, this eroded collective efficacy, changing how communities are able to pursue community development initiatives under the structural constraints of a weak postcolonial state in the periphery. In contrast, when citizens believe that the locus of power resides within the community—as opposed to with the central state—this should foster more effective local institutions through a virtuous cycle of action and legitimation. The latter case, we suggest, is most likely in contexts where state-building has emphasized local autonomy.

State-Building in the Ghana-Togo Borderlands

We test this argument in the context of the Ghana-Togo borderlands. Originally colonized by Germany, German Togoland was partitioned between the United Kingdom and France during World War I. Because of the brevity of German colonial occupation, much of Togoland was only marginally impacted by German rule—particularly in the North—and the district commissioner was oftentimes the only colonial official stationed in any given district outside of Lomé and Anecho (Knoll, 1978, pp. 43–44). In August of 1914, the British and French divided German Togoland, relying on a provisional boundary during the war that largely followed the contours of German districts. ⁴ British and French Togoland were thereafter administered as League of Nation mandate territories, though Britain administered British Togoland as part of the Gold Coast (present-day Ghana) while the French kept Togo separate from neighboring Dahomey (present-day Benin). ⁵

In both cases, Britain and France relied on the theories of colonial administration they had developed elsewhere. The British instinct toward decentralized state-building is seen clearly in the speech of the Gold Coast’s Governor in 1926, for example: “If the peoples of the Gold Coast are ever to stand by themselves, it must be by the gradual development of their own institutions and customs” (quoted in Schuerkens, 1995, p. 126). As the British grew dissatisfied with indirect rule as a development strategy post-war, the colonial state in Ghana reoriented itself by creating local governments, retaining the animating idea of a local ‘community’ as the basic building block of the state (Grischow, 2006, pp. 169–70). In contrast, French rule in Togo emphasized centralizing power. One consequence was that even though the French relied heavily on canton chiefs, by the 1930s, many chiefs had ceased being ‘traditional,’ as the French emphasized state-building over the idea of ‘indigenous tradition’ (Haan, 1993 Chpt 3). Even the 1944 Brazzaville Conference, which established territorial assemblies in France’s colonial holdings, remained framed as a vehicle for African colonial subjects to accept their eventual integration into France and focused attention on colony-wide elections (Gbedemah, 1994, p. 117).

Historians have largely documented continuation—as opposed to disruption—in these levels of centralization within the postcolonial state, particularly in the hinterlands. Writing on Northern Ghana, for example, Lentz (2006, pp. 179, 237) describes the arrival of independence as opening up the local arena to greater political initiative, building directly on the local governments created by the British in the postwar period. Already by this point, she notes, the village was a key unit of action for development, with local youth associations embracing grassroots initiatives and the language of ‘self-help.’ Similarly, scholars observe that the idea of nnoboa, an Akan word invoking a traditional form of cooperation in farming, was often invoked by the government to frame rural participation in development projects (see Brown, 1986). As described by one village chief we interviewed in northern Ghana, “Kwame Nkrumah said this is our country so we should come together do something for ourselves. He advised us to help ourselves” (Village KE4, June 12, 2019). This was crystallized perhaps most clearly in the efforts of President Jerry John Rawlings to form decentralized grassroots committees in the 1980s. Following state retrenchment, local participation became a vehicle for accessing state resources, with citizens “form[ing] new associations, organiz[ing] self-help labor and initiat[ing] claims on government (Wiemer, 2021, p. 116). Despite decades of political upheaval, therefore, the Ghanaian post-colonial state has often emphasized the importance of local political action where the state is otherwise absent.

In contrast, Piot (2010) describes the rise of participatory development in the 1990s as a sudden shift for Togolese. Prior to this, the Togolese postcolonial state was highly centralized and interventionist, even in its periphery, continuing the colonial practice of relying on chiefs to serve state interests in rural areas, such as collaborating on state projects and ‘modeling obeisant citizenship’ (29). Eyadema’s regime relied heavily on the top-down, state-run Affaires Sociales to meet its promises of accelerated development and economic growth, reflecting the highly centralized nature of state action (Nugent, 2019, p. 450; Piot, 2010, p. 139; Toulabor, 1986). Even following the shift to grassroots participation in the 1990s, Haan (1993) describes a cool embrace of participatory initiatives among the peasants of Togo’s Savanes (a peripheral region in the far North); participatory work “is never done unless the government does a new mobilization and unless it puts the necessary means at villagers’ disposal” (231). On balance, the historiography on post-colonial Togo provides little evidence that the Togolese state broke with the centralizing tendency of the French colonial regime.

While these styles of state-building broadly capture the ideological orientations of the colonial and postcolonial periods, we argue that they were, in fact, intensified in the hinterlands. Boone (2003, chapter 4) extensively documents the ways in which the postcolonial state in Ghana practiced intensely direct rule in the core regions of the cocoa belt. Under Nkrumah as well as subsequent regimes, the state went to great lengths to coopt and directly control the wealthy cocoa elite in the prosperous Ashanti region in order to extract wealth and prevent political consolidation. Where there is power and wealth at stake, namely in regions that grow cash crops, the legacies of colonial rule are less likely to endure. As a result, we limit our argument about colonial legacies to non-cash-crop growing hinterlands.

Interviews conducted in peripheral zones of the Ghana-Togo borderlands broadly accord with these differences. The chiefs and elders we interviewed in our village sample were all aware of critical differences in the nature of colonial administration on either side of the border. For example, in one northern Togolese village, the chief described French colonial rule as follows: “If someone disobeyed them, they would beat them and tell them to carry palm tree trunks as punishment” (Village TBY6, June 19, 2019). In contrast, Ghanaian villages often spoke of the British as being largely absent, “The British never troubled us here” (Village, GE2 June 13, 2019), or only belatedly involved in the community: “The time we really got involved in British administration was like the 1950s … I think it was around that time that I can say we were under the British” (Village GE11, June 12, 2019).

These differences were often crystallized by respondents in terms that neatly echo the more interventionist, top-down approach of the French. When describing the nature of the colonial experience in his village, one Togolese village elder summarized: “It was the French who dictated the orders that the chief must impose on the people” (Village TBY2, June 19, 2019). As described by another “when the French came … chiefs were not having power. What they say is what the chief will do.” (Village TK4, June 15, 2019). An inverse pattern characterized responses heard in Ghana. “They … let the people do what they want. I remember there were district commissioners in Ho and Kpando. Every month the chief sends records books to them for perusal. Sometimes they even come to the communities to see the work that the chiefs were doing. Their relationship was cordial” (Village GE8, June 11, 2019). Echoing a common tendency on both sides of the border to compare the French or British to the Germans, another chief recalled the British more negatively. In contrast to the Germans, who many recalled as having built a fair amount of infrastructure, another chief lamented “the British were forcing us to do things for ourselves …” (Village GK4, June 12, 2019).

Echoing patterns documented by Nugent (2019), impressions of the post-colonial era are similar. In particular, Wilfahrt and Letsa (2023) document that Togolese citizens are more likely to report that chiefs continue to be subordinate to state authority, in contrast to the view that Ghanaian chiefs have their own, autonomous sources of power and are oftentimes even seen as having authority over government officials. Many descriptions of relations with the colonial and postcolonial state were slippery in nature, revealing perceptions of continuity. Take one chief’s description of the relationship between chiefs and the Togolese state:

Chief: Everything comes from the Lomé ministry. The Interior Minister takes care of the chiefs and the Paramount chiefs. He will send messages to the village chiefs, or maybe they will invite all of the chiefs and tell you that this and that is going on. That is what they will tell you before you will come to tell the villagers.

In this way, Togolese chiefs described themselves as subordinate to their local préfets: even when préfets are helpful, they are in charge. Take, for example, one chief who recounted how their préfet had told him, “He would go to the government on our behalf [about the need to repair classrooms]. We are villagers, and we accept whatever we are told. We are still waiting” (Village TE12, June 15, 2019). These differences were visible across the border; one Ghanaian chief observed a difference in the simple act of cutting down a tree growing on farmland: in Ghana, he said, a farmer can simply do this as they see fit, but in Togo, state agents will fine you if you do not first obtain permission from the authorities (Village GBY1, June 20, 2019). These anecdotal differences highlight the patterns we expect to shape variation in local collective efficacy: whereas more interventionist states such as Togo seek to channel citizen developmental action towards the central state, cases like Ghana reveal a state that has been more permissive if not encouraging of developmental action that is localized and autonomous from the state.

Research Design

Empirical Analysis of the Argument

Control Variables

Table 1 presents descriptive statistics of our collective efficacy factor variable as well as a suite of control variables. As the table shows, there is some imbalance across the border. These differences are only rarely substantively meaningful, however. This is particularly true for respondent age, household SES, education level and village population. ¹¹ The most striking difference is the higher degree of ethnic diversity in Togolese villages. This difference is largely driven by villages sampled in Togo’s north, where 68% of total ethnic minorities are Bassar live in Bimoba villages. Although these groups are very close linguistically and have long co-habited, this is a notable difference, and we control for it – as well as all other unbalanced covariates - in our models. We additionally control for an individual’s level of reported self-efficacy and an index of village infrastructure. Togolese villages do have fewer goods per village, but the difference – 0.25 on an index that ranges 0 to 5 – represents only a fifth of a standard deviation. On both sides of the border, villages generally have just under three goods from our basket of a primary school, secondary school, clinic, electricity and clean water access.

OPEN IN VIEWER

Table 1.

Descriptive Statistics.

	Total sample	Ghana	Togo	Diff of means
Collective efficacy factor	3.78	3.98	3.6	0.38***
Age	40.59	38.99	42.18	−3.19**
Gender	1.5	1.51	1.49	0.02
Education a	2.51	2.70	2.33	0.37***
Household SES	0.26	0.29	0.23	0.06***
Time in village b	4.54	4.49	4.59	−0.10
Ethnic minority	0.16	0.12	0.22	−0.10***
Infrastructure index	2.82	2.84	2.69	−0.`5***
Village population	994.1	945.1	1041.3	−96.22**

***p < .001; **p < .05.

^a1 = No formal schooling; 2 = Some Primary; 3 = Primary Complete; 4 = Some Secondary; 5 = Secondary Complete; 6 = Post-Secondary.

^b1 = 6–12 months; 2 = 1–3 years; 3 = 4–7 years; 4 = 8–11 years; 5 = 12+ years.

Main Results

We employ Rabe-Hesketh & Skrondal’s (2002) joint compound hierarchical ordinal probit (CHOPIT) model to anchor the self-assessment questions described above. This method allows us to account for individuals' interpretation of the survey questions. Our primary models, presented in Table 2, report a factor variable of our ten collective efficacy questions, preserving the original answer ordering of the anchoring vignettes. As seen in the coefficients for θ₁ and θ₂, our vignettes – the first of which describes a community with high collective efficacy and the second a case of low collective efficacy – are ordered as we expect, indicating that respondents understood the concept as we intended.

OPEN IN VIEWER

Table 2.

Model Results for Collective Efficacy Factor With Anchoring Vignettes.

		Full sample				Cluster mean
		(1)		(2)		(3)
Equation	Var	Coeff	SE	Coeff	SE	Coeff	SE
β	Togo	−0.65***	(0.15)	−0.63***	(0.18)	−1.17***	(0.31)
	Age			−0.02**	(0.01)
	Education			−0.11	(0.06)
	Household SES			0.66	(0.61)
	Ethnic minority			−0.01	(0.23)
	Self-efficacy factor			−0.06	(0.04)
	Village population			0.00	(0.00)	−0.00	(0.00)
	Infrastructure index			−0.13	(0.07)	−0.28**	(0.12)
γ1 (strongly disagree -disagree)	Togo	−0.53**	(0.17)	−0.31	(0.23)
γ2 (disagree-neither)	Togo	0.05	(0.13)	−0.10	(0.18)
γ3 (neither-agree)	Togo	0.13	(0.12)	−0.09	(0.16)
γ4 (agree-strongly agree)	Togo	0.43***	(0.06)	0.44***	(0.06)
θ1	Vignette 1	2.38***	(0.17)	1.13**	(0.55)
θ2	Vignette 2	0.06	(0.13)	−1.55**	(0.55)
Ln(σ)	Log scale for vignettes	0.86	(0.05)	0.95	(0.05)
	N	2949		2820		50

***p < .001; **p < .05. Standard errors in parentheses. Results of CHOPIT models with exception of cluster mean, where sample size is too small to run. Model 3 reports results from ordered probit instead. Only our main variable of interest is reported for thresholds. Bolded values indicate primary variable of interest.

In line with our core expectation, Togolese respondents report lower levels of collective efficacy for their villages. In the full sample (model 2), the coefficient for Togo suggests that Togolese expectations of collective efficacy in their villages is approximately 0.6 points lower on a scale of 1–5, or around two-thirds of a standard deviation. The results are consistent when we look at the survey sample cluster means (model 3). ¹²

Collective Efficacy in Practice

As reported in Table 2, Ghanaian villagers, on average, report higher levels of collective efficacy. But does this extend beyond hypothetical situations? We asked our survey respondents whether, thinking back over the past year, their village had faced any problems with public resources. Togolese and Ghanaian villagers report that their villages faced a problem with a collective good in the previous year with equal frequency, as visualized in Figure 2(a). What differs, however, is the likelihood that the problem was resolved (Figure 3(b)); Ghanaians are more likely to report that the problem was fixed (60.6%) than Togolese (46%), a statistically significant difference (p < .000).OPEN IN VIEWER

OPEN IN VIEWER

Figure 3.

Table 2 Replications excluding cross-border migrants. With 90 and 95% confidence intervals. Model results can be found in Appendix F.

How are these problems solved? Across countries, respondents are most likely to report that villagers themselves solved the problem, but this number is higher among Ghanaians (see Appendix F). These endeavors were described at length. For example, in one small village in Ghana’s Volta region, a villager recalled, “We used to have a community borehole which up to sometime does not flow. The community then came together, joined resources and got a polytank” (Respondent 180326). In another Ghanaian village, a respondent reported that, “The painting on the wall of our basic school was badly faded, which made the buildings unattractive. Members of community contributed and bought paints and repainted all the buildings” (Respondent 179499). Even where the problem was ultimately fixed by the government, villagers in Ghana describe collective action used to get the government involved. As explained by one Ghanaian: “There was a time our clinic had no medicine at all to give to patients. A committee was set up by the community, and they came together and wrote a petition to GHS [Ghana Health Services] and later drugs were sent down” (Respondent 179390).

In our semi-structured, qualitative interviews, Ghanaian chiefs and village elders echoed the stories told by their subjects. A chief in Ghana’s Volta Region explained, “Our post office...was built by communal labor, and the contribution of our people themselves. But the schools too in the beginning were built by the people themselves…So, at times we do it ourselves. At [other] times when someone comes to us to help us, give us materials, we do it ourselves” (Village GE7, June 8, 2019). Although village chiefs often play a coordinating role for collective efficacy, they do not appear to be necessary actors for the process of collective efficacy, as one Ghanaian elder in the north described: “The women, they are most affected by the water. So, they have some small contributions. They are putting it there in case something spoils, then they will use that money to buy and they will fix it for themselves…[only] If it is beyond their control, [will] they will call on others to support [them]” (Village GBY1, June 20, 2019). Ultimately, many saw the question of addressing local issues as the community’s responsibility: faced with a local development challenge, one Ghanaian chief summarized “it is the community that will take responsibility. Even if the government will come and do maintenance, it is the community that will lead the way. If the community does not take the lead, the government will not step in” (Village GE8, June 11, 2019).

It is more difficult to present systematic evidence of a lack of collective efficacy, but Togolese chiefs and elders were more likely to recall stories of frustration, especially when it came to mobilizing labor or collecting donations from villagers. For example, one Togolese chief in the south told us, “Nowadays, people don’t participate very much [in communal labor] like before…to say that there is a village which is massively interested in community work, there is none” (Village TE1, June 16, 2019). There was certainly variation in the reports provided by the chiefs; some Togolese communities reported high levels of collective efficacy, but there was much more variation in Togo than there was in Ghana. Further, when things got done in Togolese communities, it was often through the initiative of the chief himself. For example, when asked what would happen if there was a natural disaster, an elder from a southern village in Togo replied:

Elder: We have committees who deal with these things and take action. The CVDs [village development committees] and the chief are the main culprits. Currently, it is the regent who is in place of the chief, and with the CVD members, they would walk around the village.

These differences can be quantified from our interviews with chiefs and village elders. We asked about the presence of a battery of basic infrastructure - clinics, primary and secondary schools, improved water sources, electricity and roads - in their villages. For all public goods that were present, we asked who had first constructed the good. In both countries, the least common answer was the state (24% in Ghana versus 26% in Togo). Strikingly, 47% of Ghanaian interviewees reported that they themselves had first built their public services, in contrast to only 29% of Togolese. The remainder of projects were identified as having been initiated by an NGO or church (44% in Togo, 29% in Ghana). ¹³

Despite the comparably low rates of government initiative, Togolese communities were also more likely to describe their frustrations in contacting the government for development assistance, and their collective helplessness in the face of disappointment. Despite this, the view persists that the government should lead action on resolving many local development problems. For example, in a different community in southern Togo, an elder reported that if there were winds that destroyed buildings in the community,

“it is the government, specifically Social Affairs, which must take care of it. But when you call on them, they take the money without doing the work…The wind that ripped off our roof, for example, we can’t do anything, it comes under the responsibility of the Social Affairs people. If you go to see them, they rush to come and take stock of the damage, collect the little money you have, and they leave without coming back to do the work. It's the same all the time. Recently, a case happened again. Someone’s roof came off and they came to take the money without coming back until now” (Village TE7, June 12, 2019).

Togolese communities appear to be more likely than Ghanaian ones to view the government as the final solution to many of their local problems, meaning that in the wake of the state’s inaction, problems go unresolved. In contrast, when similar frustrations with the government are reported in Ghana, villagers tended to rally together to find solutions. Chiefs generally reported that villagers were willing to provide labor as well as money to help with projects, often describing their villagers as being highly ‘cooperative.’ For example, when we asked a chief in southern Ghana how willing villagers were to participate in collective labor, he replied, “They are very happy to contribute to development. Right now, we want to build a community center, so the people of the community are bringing gravel to the site to help in the process...It is not government that is doing anything for us. Even if we want government to help us, I know it will take forever to realize that. So, we are doing that by ourselves” (Village GE5, June 11, 2019). In sum, villagers don’t just imagine their collective efficacy, they also readily describe real scenarios of collective action that happened in their villages.

Importantly, villages whose residents perceive higher rates of collective efficacy also report being more satisfied with their villages as a place to live (see Appendix D); although we cannot pin down the causal direction of this relationship, it is suggestive of a virtuous circle. On the balance, our data, both from the survey and our interviews with village chiefs and elders, support our core theoretical prediction that communities that were exposed to less intensive modes of state-building report high degrees of collective efficacy and demonstrate a greater capacity to resolve local development challenges. ¹⁴ To be clear, our claim is not that Togolese villages have no collective efficacy – we heard many stories of Togolese communities working hard to address pressing issues. ¹⁵ Instead, we are drawing attention to important variations in the nature of the African state: where the state has emphasized the center’s authority, communities are more likely to be oriented toward the state today. We see this in Togo, where village chiefs were far more likely to note that they would report any issues in their village to state authorities (generally, the prefet or chef de canton). One consequence, we suggest, is that these same communities may be less efficacious in finding community-led solutions.

Assessing Alternative Explanations

Assessing Alternative Mechanisms

State-building is an inherently bundled treatment. As we invoke the term here, we refer to more or less interventionist tendencies within the state which orient political action toward or away from local communities. We recognize, however, that other dimensions of the central state could impact levels of collective efficacy. Our qualitative data concentrates primarily on the tendency of centralized administration to focus local needs and demands upwards towards the state, which we argue is likely to undermine local collective efficacy. However, other prominent mechanisms in the literature relate to other legacies of British versus French colonial rule, for example the empowerment of local traditional leaders as well as a cultural legacies, wherein indirect rule heightened ethnic identification (e.g., McNamee, 2019).

We present tests for the chieftaincy pathways in Table 3. ¹⁶ We use three variables to capture the strength of traditional authorities: first, how frequently a respondent reports having contact with their local chief in the past year (1–5, never to daily), how likely the respondent thinks their chief would be to help with a problem pertaining to a public good in their village (1–4, not at all likely to very likely) and whether a respondent considers their chief’s opinion important when deciding which party to vote for (1–4, not at all important to very important).

OPEN IN VIEWER

Table 3.

Assessing the Relationship Between Residing in Togo and Collective Efficacy Controlling for Potential Alternative Mechanism: Strength of Traditional Chieftaincy.

	Panel (a)			Panel (b)
	Contact chief	Chief would help	Chief influences vote	DV = collective efficacy
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Togo	0.101	−0.292**	−0.257	−0.783***	−0.599***	−0.741***	−0.638***	−0.746***	−0.579**
	(0.208)	(0.097)	(0.147)	(0.190)	(0.188)	(0.214)	(0.196)	(0.204)	(0.194)
Contact chief				0.153**	−0.041
				(0.050)	(0.067)
Chief would help						0.061	−0.062
						(0.079)	(0.123)
Chief influences vote								0.005	−0.137
								(0.069)	(0.085)
OLS	Y	Y	Y	Y	N	Y	N	Y	N
CHOPIT	N	N	N	N	Y	N	Y	N	Y
Controls	Y	Y	Y	Y	Y	Y	Y	Y	Y
N	948	939	931	908	2797	903	2774	896	2753

***p < .001; **p < .05. Standard errors in parentheses. Controls are identical to those used in main models: age, education, household SES, individual self-efficacy and village population.

As suggested by Judd and Kenny (1981) and implemented by McNamee (2019), measures of the mechanisms should be both significantly associated with our core independent variable (residing in Togo), and, secondly, controlling for these measures should weaken the effect of residing in Togo. We test the first stage in Table 3a, which shows that only two of the measures of traditional authority correlate significantly with residing in Togo. As can be seen in Table 3b, in no case does the inclusion of these potential mediators dramatically influence the coefficient on Togo. In comparison to a coefficient of −0.75 in an OLS model and −0.64 in the CHOPIT models (both with controls) when the potential mediators are not included, we see little substantive changes, suggesting that an alternative mechanism rooted in the strength of the chieftaincy is unlikely to drive the effect of residing in Togo on colletive efficacy.

Perhaps the most prominent alternative mechanisms are derived from the different regime trajectories over the past two decades following Ghana’s political liberalization. We note two factors that we think limit the impact of Ghana’s democratization. First, both countries have a long history of authoritarian rule. Indeed, throughout the 1970s and 1980s, scholars pointed out Ghana’s political instability and interminable economic crises in contrast to Togo’s “relative stability and prosperity” (Bening, 1983, p. 207). ¹⁷ Because we are interested in the local dynamics of community problem solving, we further flag that in both cases, local governments – district assemblies in Ghana and prefectures in Togo – are run by central state appointees. This means that the most immediate level of government for villagers on either side of the border is more accountable to the center than to local interests. Because districts are tasked with implementing state development plans, this is not inconsequential for how communities pursue development activities. We discuss this in more detail in Appendix F. Finally, even within democratic Ghana, the areas sampled by our survey have traditionally been a stronghold of support for the NDC party, where it consistently wins landslide victories in local elections, particularly in the South, making for little competition during elections. Thus, in terms of local-level political competition along the border, Ghana and Togo may be more similar than current national-level democracy scores would imply.

Nonetheless, in order to assess the plausibility of this alternative explanation, we asked respondents a question specifically aimed at eliciting feedback on whether they felt collective efficacy had declined in their village and why. We did so for two time periods: the 1980s and 1960s. ¹⁸ If collective efficacy is driven by democracy rather than modes of state-building, we would expect Ghanaians to perceive collective efficacy to have increased in recent times (and Togolese to report no change). In contrast, or both the 1960s and 1980s, Togolese respondents are significantly more likely to report that, if anything, collective efficacy has increased in their villages since those periods, while Ghanaians believe it has decreased. The distribution of these answers can be found in Figure 4. Note that these results hold even when we restrict the sample to those who were adults during these decades and we similarly do not find evidence of strong cohort effects in Appendix G.OPEN IN VIEWER

Figure 4.

Perceived changes in collective efficacy over time.

Generalizability

Conclusion

This paper has provided evidence that different modes of state-building have impacted collective efficacy of rural communities in West Africa, and that, in turn, this variation has important consequences on development outcomes in these communities. Controlling for a host of factors, including socioeconomic status, education, and even individual efficacy, we document that villages in Ghana systematically possess higher levels of collective efficacy than villages just a few kilometers away in Togo. Ghanaian communities perceive themselves as being better able to come together to improve their villages, both hypothetically and when they report actual instances of collective action in the past year. In contrast to existing work that measures capacity for collective action as a function of ethnic homogeneity, our findings suggest that collective action is better theorized as a property of institutional legacies. Where the state has actively sought to diminish the power of local communities in favor of the central state, communities appear to possess lower levels of collective efficacy.

Even though those we spoke to for this project emphasized durable social connections with their neighbors across the border, with whom they shared deep cultural and economic ties, exposure to the colonial and post-colonial state had left clear marks. Such legacies of distinct approaches to state-building may be particularly sharp in border regions (Braun and Kienitz 2022; Nugent, 2019), but the evidence presented in this paper emphasizes a clear need to pay attention to how distinct histories of state-building have generated unique political cultures, even in the face of over-arching state-weakness.

This has important implications for some of the poorest communities in the world. West African communities are often expected to coproduce public goods and the ability to organize to fix of a borehole or build a classroom means the difference of having access to potable water or an education. This makes understanding the ability of communities to overcome collective action problems a paramount concern. We have suggested that collective efficacy is in part shaped by variation in the nature of state-building - where the colonial and post-colonial state has emphasized the central state’s authority, communities remain oriented toward the state today and are less efficacious in finding community-led solutions. Other historical experiences likely also matter. Although the communities we worked in have not experienced notable levels of violence or contention, future research could address when violence dampens collective efficacy and when it might enhance it by providing communities with experience in solving communal problems. Whatever the nature of this variation, the idea that communities may be differentially equipped to solve local development problems reveals a core tension that emerges from our findings: as the push for participatory initiatives continues at the same time that post-colonial states back away from their obligations to rural citizens, disadvantaged communities come to bear more of the heavy burden of development and public goods maintenance, which risks reinforcing, if not exacerbating, these inequalities.

Supplemental Material