In this chapter, I turn to an original dataset of historical public service investments to test a final implication of my theory. While the previous two chapters demonstrated that congruent local governments engage in spatially distinct redistributive politics, my argument holds uneven temporal predictions: the influence of the precolonial past on redistributive politics should be contingent on its congruence with formal institutions. As I show in the following pages, the precolonial past only influences social service delivery following the 1996 decentralization reforms that transferred authority over public goods placement to local governments.
Built from archival documents, my historical dataset also enables me to assess a set of possible alternative explanations first raised in Chapter 2. The colonial era brought numerous changes to rural Senegal that have been shown elsewhere to radically alter development trajectories. Could the findings I document be driven by divergent experiences with the colonial state and not exposure to precolonial centralization? I test a range of arguments about colonial influences on long-run development trajectories but find little evidence that colonial legacies mediate those of the precolonial past. Thus, while French colonization did intimately influence the contours of Senegal’s social service infrastructure, the political and economic interests of the colonial state only appear to have shaped proximity to basic services in the colonial and immediate postcolonial period. Colonial effects have almost entirely faded by the early 2000s.
In this way, this chapter “decompresses” history to engage in two broader debates animating the recent turn to history among students of political economy of development. By explicitly examining the impact of precolonial and colonial legacies over time, I offer a corrective to the tendency to gloss over the question of temporal process that defines much of our recent interest in historical legacies. As the empirical evidence marshaled in the following pages shows, I find no support for arguments that areas that were home to precolonial states are doing better today because they have always done better. In sharp contrast, my data reveal that legacies can persist at the grassroots only to reemerge to influence the outcome of interest under specific institutional arrangements.
From here, I draw out a second, related tension in much of the recent scholarship on historical legacies. This work has largely assumed that historical legacies are long-run equilibria, generating the expectation that they should persist absent some exogenous shock. Yet as I document, empirical trends in the legacy of colonial service delivery seem better suited to evolutionary logics that are more consistent with the changing political dilemmas of the colonial and postcolonial African state.Footnote 1 Without taking temporal logics seriously, I caution that we risk making inaccurate predications about how and why the past influences the present.
The chapter is structured in three parts. I begin by introducing the historical data, before establishing that the effect of falling within the territory of a precolonial polity only becomes consequential for public goods delivery following the 1996 decentralization reforms. I secondly examine whether this obscures or ignores an important role for colonial legacies, only to find that these effects have largely faded by the 1990s. A third and final section develops the implications that these findings hold for research on historical persistence in the study of political economy of development.
When Does the Precolonial Past Matter?
In identifying a robust legacy for West Africa’s precolonial kingdoms, this book contributes to the increasing recognition that African centralized precolonial organizations have long-lasting impacts on development outcomes.Footnote 2 Prominent arguments in this literature suggest that areas that were home to precolonial states have better living standards today because they were home to distinct cultural norms, more accountable chiefs or because they were better able to adopt new technologies to name a few examples.Footnote 3 Like most work on historical legacies, the overarching assumption among these arguments is that something about the precolonial past matters, that it has consistently done so, and that its impact is cumulative.
My own argument departs from this linear causal story: precolonial states left behind robust cross-village social institutions that today generate distinct forms of distributional politics. Social institutions have not always mattered for development, however, because their ability to shape public goods delivery is contingent on the formal institutional configuration that they operate within. Key for my argument is that the 1996 decentralization reforms were an exogenous shock that facilitated the (re)emergence of social institutions as a key driver of subnational distributional politics. Accordingly, I examine whether we see evidence of interactive effects between precolonial legacies and formal institutional structure, only to find that precolonial legacies do not become a significant factor until the 1996 decentralization reforms.
Historical Data
To trace the impact of precolonial political centralization on development outcomes, I use archival data to extend my village-level dataset back to 1882, coding the construction of new primary schools and basic health facilities in rural Senegal at roughly ten-year intervals to the present.Footnote 4 To do so, I rely on annual colonial reports by sector, notably the Rapports Statistique Annuel de Santé and d’Enseignement, accessed at the French Archives nationales d’outre-mer (ANOM) or the Archives nationales du Sénégal (ANS) in Dakar.Footnote 5 For the first decades of colonization, the data are completed with the Annuaire du Senegal et Dependences.Footnote 6 This allows me to generate a dummy variable of whether or not a village has access to a primary school or a health facility, which may include a health center or dispensarie (“poste” post-1975), in a given year using the same definition of access introduced in Chapter 5. Health huts (cases de sante), more basic health facilities administered under health posts, were not pioneered until the early 1980s and are regrettably not included in the dataset due to a lack of information about their locations prior to 2000. Unsurprisingly given the state crisis of the era, village-level data are not available between the mid-1970s and 2000, leading me to complement the village-level data with alternative measures that, though aggregated at a higher level, help fill this temporal gap. These are introduced below.
Because many villages were settled following the onset of colonial rule, I restrict my analysis for the early years of colonialism (1902–1912) to villages (n = 3,301) that are listed in the first censuses conducted by the French between 1890 and 1904.Footnote 7 Early census data could not be located for some parts of the country during this time period and missing zones span the country’s precolonial political geography. Notably, acephalous zones of Kedougou Region and the Ferlo Desert are missing, as well as present-day Kolda Region, which had a mixed political history, and Tivaouane Department, which was home to the heart of the precolonial Cayor Kingdom. I use data from the 1956 Repertoire de villages to expand the scope of villages in existence for the 1922–1972 period (n = 10,877).Footnote 8 Restricting the data in this manner helps mitigate the risk of estimating the nondelivery of public goods to villages that were not in existence and hence not eligible for selection, though of course using reduced samples is an imperfect solution because I am unable to precisely estimate when villages entered the sample.Footnote 9 Due to a lack of digitalization for the archival records, all data were hand-matched.Footnote 10
Estimating the Legacy of Precolonial Statehood
I begin by assessing the ability of precolonial centralization to explain a village’s likelihood of receiving school or clinic access over time. Figure 7.1 shows the effect of one standard deviation shift in institutional congruence on whether a village gains new social service access (a primary school and/or health clinic) at approximately twenty-year intervals from 1902 through 2012 from a series of naive models. As is abundantly clear, the effect of precolonial centralization on social service access in any given village is only statistically significant post-2002. There is no significant effect in earlier eras. As late as 1972 the effect is actually negative, suggesting that historically uncentralized areas received more investments from the late colonial and early postcolonial states.
Effect of institutional congruence on new social service access over time
Though suggestive, missing periods of observation in the village-level data limit my ability to pinpoint the mid-1990s as a pivotal moment. More complete time series can be found at the department level, a higher level of aggregation for which ministerial data are available for the period missing in Figure 7.1. Colonial-era data are aggregated upward to these boundaries. Figure 7.2a shows the average number of schools by department, with diamonds indicating the years for which data are available. The figure illustrates a marked flip between the average number of schools in departments that were dominated by a precolonial state and those that were not following the 1996 administrative reforms. While historically acephalous areas had more schools on average from the 1940s through the early years of independence, this effect switches dramatically following the devolution of authority over primary school construction to local governments.Footnote 11 Between 1998 and 2002 for example, departments that were dominated by precolonial states areas outbuilt acephalous ones by 2.7 schools to 1. Between 2002 and 2012, this ratio was 1.8 to 1.
Basic services over time: (a) primary education, (b) basic health facilities, and (c) boreholes
Figure 7.2b replicates Figure 7.2a for the average number of health posts by department. Here areas that were home to centralized states in the precolonial era have slightly more health posts on average beginning in the 1940s (though the difference is small at less than one post per department until 1976). While there is no switch as was seen with primary schools, the growth rate in average health post construction following 1998 is accelerating at a faster pace in departments whose territory housed a precolonial state than those that did not by roughly two to one.
To reinforce my argument that these changes in building patterns are unique to locally delivered public goods, Figure 7.2c shows comparable data for borehole construction. Major waterworks like boreholes are equally if not more in demand by rural populations, yet this is not a devolved competence and such projects are far beyond the capacity of even the most well-financed local governments. Borehole construction has always been in the control of the central government and, as Figure 7.2c illustrates, it has been biased toward centralized areas from the postcolonial period on; most of this difference is driven by investments in Senegal’s peanut basin. Still the data suggest that if anything there has been a reduction in this bias in the 2000s as historically acephalous zones have received more boreholes relative to historically centralized areas than in the post-decentralization era. Central-state-provided goods seem to behave more consistently over time than locally provided goods, which see a visible change in patterns of their delivery following decentralization.
As a final means to address data limitations, Figure 7.3 shows the marginal effect of centralization on educational outcomes as reported by rural respondents in the 1992 and 2010 Demographic and Health Surveys (DHS) for Senegal. The DHS surveys are nationally stratified surveys, such that households have equal probability of being sampled within each census enumeration zones of a given country. The surveys collect a complete inventory of household members as well as their individual demographic attributes. Given the DHS’s wide sampling of respondents across the country both before and after decentralization, I am able to estimate differences in “stocks” of education in rural residents as a means to establish the availability of public goods across space and time.Footnote 12 The sample is truncated to those who would have been primary school aged in 1996 and onward; hence the 2010 data are limited to those between the ages of fourteen and twenty-one. To ensure comparability, the 1992 data are limited similarly. I estimate health outcomes by looking at the percent of births per woman over the previous five years that took place in a public clinic. The data tell a now-familiar story: young rural Senegalese were less likely to be educated if they lived in areas that were home to centralized states in 1992, but this gap has closed by 2010, reflecting the gains in primary school investments made over this period. Similarly, women living in areas that were home to a precolonial state were significantly less likely to have delivered their youngest babies in a clinic in the early 1990s, but the effect flips direction by 2010 when they are more likely to deliver in a clinic. Importantly, there is no reason to expect that this would be driven by cultural shifts. It is hard to imagine why attitudes about Western education or, in particular, health care, would only accrue in historically centralized areas and to a significant enough degree to generate shifts of this scale.
Marginal effect of centralization on education and health attainment
What Precolonial Centralization’s Intermittent Effects Say about Historical Legacies?
Figures 7.1–7.3 cumulatively reject the argument that precolonial centralization has been a constant force for improving access to public goods. I find no support for the idea that precolonial polities disproportionately attracted colonial investments, thus that I would conflate a colonial for a precolonial legacy. This does not appear to be specific to Senegal. Ricart-Huguet (Reference Ricart-Huguet2021) likewise finds little impact of precolonial statehood in driving district-level colonial investments across sixteen West and East African colonies, suggesting that precolonial statehood did not robustly influence colonial investment strategies.
The sudden significance of precolonial centralization seen in Figures 7.1–7.3 is not easily explained by anything other than Senegal’s decentralization reforms. Still, I note here three other significant changes took place in this general time frame that may strike some as plausible alternatives. As I discuss, however, none are likely drivers of the shifts observed in the data.
First, we might think that structural adjustment reforms, adopted widely across the continent in the 1980s, may have had uneven subnational effects.Footnote 13 Certainly, structural adjustment programs encouraged governments to scale back spending. To illustrate, Senegal’s health spending – which averaged 8–9 percent of the national budget in the early years of independence – had dropped to 5.4 percent by 1985, evidence of the drastic cuts that took place during the early years of structural adjustment.Footnote 14 Ultimately, however, the neoliberal reforms of the era did little to fundamentally alter the structure of service expansion, having its most focused impact on central government bureaucracies on the one hand and, on the other, end-line users through the introduction of cost recovery programs in the late 1980s.Footnote 15 There is no evidence that structural adjustment altered the clearest distributional pattern of the postcolonial state: a durable bias toward urban service investments.Footnote 16
A second notable change was the significant uptick in development funding, specifically focused on basic services. By the mid-1990s, only 58 percent of the eligible student population attended primary school and only 27 percent of adults were literate, far below the continent’s average which hovered around 50 percent.Footnote 17 The international community sought to address such disparities with renewed vigor from the 1990s onward. This means that Figures 7.1–7.3 may reflect the ability of the state to politically target certain populations with new donor funding rather than these effects being driven by local social dynamics.Footnote 18 If this were the case, however, it is unclear why the regime would engage in inconsistent patterns across goods, favoring historically centralized areas with primary schools and clinics while favoring historically acephalous zones with high-demand goods like boreholes.
The election of Abdoulaye Wade in 2000 – the country’s first electoral transfer of power that saw the defeat of the PS – is a third and related important change during this time period. How do I know that it was the 1996 decentralization reforms introduced under President Abdou Diouf and not his loss to Wade four years later that drives these distributional changes? Though I cannot establish this with absolute certainty, Wade’s regime is not considered to have radically altered Diouf’s general political orientation. Indeed, on many metrics their regimes were very similar: both ruled through rural clientelist networks, reinforced the role of the presidency, and co-opted challengers.Footnote 19 This is not to deny their differences. Though both regimes relied on tight alliances with Sufi leaders as vote brokers, Wade’s regime was notable for its shift toward Touba, the seat of the Mouride Brotherhood, at the expense of the Tidjanes, based in Tivaouane.Footnote 20 Still, both men represented what Kelly (Reference Kelly2020) has called “insider-induced turnovers.” Wade served in Diouf’s government in the early 1990s and had long been involved in Senegalese politics, meaning that while his election was monumental, the 2000 alternation of power largely did little to challenge the constellation of national elites as Wade continued Diouf’s strategy of encouraging the formation of patronage-seeking parties that the regime could co-opt. Analysis of the two men’s regimes offers little reason to think that Wade systematically shifted service delivery toward regions of the country that had been home to precolonial states.
Rather, I argue that the devolution of authority over the construction of primary school and basic health facilities to local governments that came with Acte II of Senegal’s decentralization project offers the most leverage on the changing patterns we see in the data. These reforms were arguably the most significant change to the structure of primary education and health delivery in postcolonial Senegal. Foley (Reference Foley2010, 58) corroborates this, writing that the local officials she interviewed in the late 1990s saw Acte II as much more significant than the previous decade’s neoliberal adjustment given the significant administrative restructuring of responsibility that it brought to local actors. Beyond creating new political space locally, Acte II also nuanced the authority of long-standing patronage structures, most notably the role of the Muslim Brotherhoods as key brokers between the state and rural communities.Footnote 21 This is widely remarked upon in my own interviews as well. One former mayor described how no one in his community was very interested in the first elections in his local government, held in 1976, because the local government was seen as largely irrelevant. It was only after 1996 that the availability of real resources garnered the attention of local elites who were attracted to the new authority granted to the local state.Footnote 22
This is an important corrective to recent work on precolonial legacies, which tends to assume an equilibrium dynamic. Much of this literature invokes “catch-all” mechanisms that are unable to explain variation in the antecedent cause’s effects over time. For example, Hjort (Reference Hjort2010, 689) argues that centralization among the Tswana perpetuated norms of individual property rights, inter-ethnic unity, and democratic traditions as an explanation for Botswana’s postcolonial economic growth. In assuming that these mechanisms are constant over time, however, Hjort and others who find similar legacies cannot explain why we might expect a positive effect at some points in time but not others. If Senegal’s precolonial states similarly left behind, say, a legacy of inter-ethnic unity, why wouldn’t this have impacted state redistribution from at least independence onward? This question is important because one clear measure of an argument’s strength is the plausibility of its mechanisms as a link between the outcome and the antecedent condition.Footnote 23 This necessitates taking historical processes seriously and developing mechanisms that are consistent with temporal processes. Doing so reinforces my theory of institutional congruence: the legacy of the precolonial past is dependent on formal institutional arrangements. It is only when formal and informal institutional boundaries are congruent that we see an effect on precolonial statehood.
Assessing Alternative Historical Legacies: What Role for the Colonial State?
Above, I document little evidence that the colonial state targeted investments to the territory of Senegal’s precolonial kingdoms. In view of the stunning reinvigoration of academic interest in the impact of colonization on development outcomes, however, colonial legacies may not be so easily dismissed.Footnote 24 Increasingly, this has led scholars to turn their attention to subnational variation in colonial rule, enabling us to nuance our understanding of how colonialism matters by revealing how colonization was a spatially and temporally uneven exercise in state-building.Footnote 25 As a result, causal primacy has been ascribed to the colonial state in lieu of precolonial or postcolonial legacies. Although some scholars have begun efforts to adjudicate between seemingly contradictory conclusions, the key tenet of this literature continues to be that exposure to some aspect of colonization (for instance, specific institutions or forms of human capital transfer) generated path-dependent effects that persist to the present.Footnote 26 In this view, contemporary conditions in the Global South are a function of a series of colonialism-induced equilibria.
For many, this will make my claim that it is precolonial legacies that are influential in the present puzzling. Could the effects I ascribe to precolonial political geography be masking what are ultimately colonial-era changes? To account for the possibility that my data may obscure legacies induced by the distributional logics of the colonial state, I test three common claims about the impact of the colonial encounter below. First, I examine whether contemporary development is a function of location fundamentals. This logic would predict that patterns of social service access were determined by proximity to the colonial state. A second and related argument is that service delivery is spatially “sticky,” with early exposure to social services predicting long-term access. Finally, I address a more diffuse set of arguments that the colonial state delivered services to fulfill its own specific political agenda with lasting impacts.
Did Early Colonial Exposure Drive Outcomes?
Early colonial investments were thin and clustered near points of French strategic investments. Prior to French expansion inland, the first schools and health facilities were founded in the ports and escales dominated by French traders, principally based in and around the early urban centers of Saint-Louis and Dakar. Education slowly expanded outward on three axes: in an expanding network of trading centers as French commercial interests spread along the Senegal River and, later, along the Saint-Louis–Dakar railroad constructed in the late 1880s; via missionary stations along the coast; and in the early administrative posts of the young colonial states as French officers made tentative efforts to provide primary health and education services. Health facilities were much slower to expand beyond major population centers and much of the country, even areas that had seen relative advances in education, remained drastically underserved. Neither rural Diourbel nor Fleuve (present-day Saint-Louis and Matam) Regions had a doctor prior to 1910, for example.
In part, this reflected prevailing beliefs in Europe at the time that economic and social activities were the responsibility of the private realm with the state focused on providing internal and external protection. While over time it became clear that the colonies would never attract private investment without some state-led development of infrastructure, such as railroads and basic social services, the constraints of colonial budgeting, which required colonies to be economically self-sufficient hampered significant development efforts.Footnote 27 Early revenue raised in French West Africa was largely destined to finance the colony’s own administration.Footnote 28
The consequence was that while the colony’s small European populations did have access to basic social services, most colonial subjects effectively received nothing in terms of investment from the colonial state in the first decades of colonization. What impact was felt in outlying areas was the expansion of transportation and communication networks, the introduction of a uniform currency, and the alienation of peasants from traditional land-holding practices.Footnote 29 Though many colonial administrators bemoaned the lack of development effort on the part of the state, with high-profile actors like Gouverneur Général Ponty calling for greater investment in the domains of health and education in 1908, efforts remained weak. From the 190 primary schools in operation in French West Africa in 1909, there was no meaningful increase by the end of First World War a decade later.Footnote 30
What public goods were built were often done so on the whim of individual administrators, who had absolute authority in their cercles far from the oversight of their superiors.Footnote 31 In contrast to the dominant view of the French colonial administration as a centralized, homogenous entity, in reality, the early years of colonization are best characterized as a “decentralized administrative structure [that] virtually gave a free hand to the man who wanted to build a bridge, establish a schoolhouse, or help increase local peanut production ….”Footnote 32 The French lacked a coherent vision for their massive territorial acquisitions, which granted substantial authority to the understaffed territorial administration, though this also meant that any given official’s individual efforts were largely circumscribed to the most accessible areas around his post and easily abandoned upon his departure.
The only coherent education policy in the early years of colonial administration in French West Africa was a desire to create and reproduce a stratum of Africans able to perform vital functions for the colonial economy: government clerks, technicians, veterinarians, teachers, and doctors.Footnote 33 Following the 1912 request of the government of French West Africa government that each colony create its own Education Department (Senegal’s was created in 1913), schools were organized in a pyramidal fashion: village schools at the base, regional schools serving as a middle layer, and a few elite schools, based in Dakar and Saint-Louis, that trained future elites at the top.Footnote 34 What schools were built in rural districts often suffered from weak attendance.Footnote 35 Early health intervention was simultaneously geared toward meeting colonial needs and structured along racial and class lines.Footnote 36 Although the French made significant advances in expanding access to health care relative to other colonial powers on the continent, in large part because they invested in training African doctors, clinics remained concentrated in administrative centers much like schools; the average distance to a clinic in the 1920s, for example, was just shy of 50 kilometers.Footnote 37
The clustered nature of early colonial investments raises the possibility that social service access was merely driven by the happenstance of proximity: being located near a European population center, a trading post, or an administrative center with an ambitious colonial officer were all likely to increase access to services. Alternatively, if exposure to precolonial coastal trade influenced district-level colonial investments or if subsequent investments diffused outward from early missionary posts, for example, we might find certain path dependencies in service access.Footnote 38 Given that the early colonial state made few investments, am I actually capturing persistent “spatial equilibrium” that is truly a function of early proximity to colonial infrastructure?Footnote 39
To answer this question, I turn again to my dataset of historical public goods delivery. I estimate the effect of proximity to the colonial state in four ways. First, I look at whether distances to European population centers can explain social service access over time. I measure this with (a) a village’s logged distance to the nearest French outpost – fort, mission, trading post, or rail station – before the full onset of colonial rule (pre-1890) and (b) a locality’s European population in 1912.Footnote 40 I secondly measure exposure to the colonial administration with the logged distance to the nearest French administrative center in 1902 and again in 1922, when the colonial state was more entrenched. Third, I take into account transport infrastructure by looking at the logged distance to the nearest point on the French railway in 1910, when it was largely completed, and to a major road as of 1940, when road travel had come to surpass rail in much of the countryside. Lastly, I examine exposure to missionaries with the logged distance to the nearest mission in 1900 and 1918, after which the map of missionary activity was largely fixed in rural areas.
Results from a series of OLS estimates are displayed in Figure 7.4. These models control for the same geographic variables introduced in Chapter 5 because geographic factors may have both encouraged French settlement or investment decisions while also influencing subsequent development processes. Models are run with fixed effects at the cercle, an early administrative unit that is pegged here to its 1915 boundaries, the first year for which complete French colonial maps are available. While choosing a geographical boundary that was created following colonization is far from ideal, these boundaries are arguably less politically informed than subsequent administrative divisions because early efforts to delineate the colony quickly amalgamated large swaths of land as the French moved inland. All models include robust, clustered standard errors at this level as well.
Effect of colonial exposure on service access over time: (a) proximity to Europeans, (b) proximity to French administration, (c) proximity to rail and road, and (d) proximity to missions
Two things are immediately obvious. First, proximity to the colonial administration and Europeans – via missions or those living near French administrative centers – is almost always significant in the early colonial period. Likewise, areas that had earlier access to road and rail networks see similar boosts in their exposure to services during the colonial era. This suggests that early points of accessibility into the rural countryside did catalyze the delivery of social services. The second conclusion that we can draw is that while these effects are remarkably sticky through the end of the twentieth century, they disappear entirely by the start of the second decade of decentralized governance if not earlier. Location fundamentals induced by the colonial state do not appear to be driving the distributional patterns documented in earlier chapters of this book.
Are Colonial Development Policies Durable?
A variant of the argument explored above is found in Huillery’s (Reference Huillery2009) paper that argues that infrastructural investments, such as clinics and schools, are sticky: colonial investments tend to produce persistent patterns of access over time. If, as Huillery (Reference Huillery2009) argues, colonial-era investments drive subsequent ones, then contemporary patterns of access may simply reflect economies of scales in areas that received higher levels of investment during French colonial rule.Footnote 41 To evaluate whether this is the case, I examine whether access to services during the colonial era can predict subsequent access.
Figure 7.5 shows the line of fit between a village’s distances to the nearest school (Figure 7.5a) and health center (Figure 7.5b) in 1912 and at four other points in time: 1932, 1952, 1972, and 2012. Panel A suggests a positive correlation between distances to education facilities in 1912 and 1932, but this relationship flattens by the late colonial era and remains so to the present. Panel B shows that healthcare access correlates more strongly over time, though this relationship has also weakened considerably by the early 2000s.
Linear fit between early colonial and subsequent service access.
Education.
Health.
The flattening in Panel A reflects in part the uptick in colonial investments following the Second World War. At the 1944 Brazzaville Conference, convoked by French President Charles de Gaulle to outline a plan for the African colonies following France’s liberation from Germany, a massive expansion of education and health services was proposed, motivated by the belief that the colonial state was “building Frenchmen.”Footnote 42 During the same year, the Governor General of the AOF commissioned a fifty-year plan to ensure mass education of African subjects that resulted in primary education reform in August of 1945, whereby colonial schools adopted the standards of metropolitan France.
The ten-year plan for the Modernization of Overseas Territories of 1946 aimed to double school enrollment in five years and triple it by 1956 to meet an overall goal of 50 percent enrollment throughout West Africa. In 1948, primary school enrollment was opened to all.Footnote 43 Although the region fell well short of these numbers, more than 1,000 schools were built in French West Africa between 1949 and 1950 through the Fonds d’Investissement pour le Developpement Economique et Social des Territoires d’outre mer (FIDES).Footnote 44 FIDES brought not only money, but technicians, substantially restricting the latitude previously afforded administrators as service delivery was increasingly bureaucratized.Footnote 45 Perhaps most notably, the FIDES era saw significant expansion into areas that had previously seen little investment as the colonial state, now more secure in its hegemony, began extending a more comprehensive reach across the countryside. These gains were meaningful but insufficient.Footnote 46 While the budget for primary education tripled from 1,595,000 CFA to 5,311,000 between 1950 and 1955, the impact on student enrollment remained low: with 4 out of 1,000 students educated in 1947, this number only increased to 10 in 1,000 by 1957.Footnote 47
The significant expansion of services during the FIDES era calls into question the assumption that it was the colonial states’ earliest investments that drive long-term access. I test to see how early versus late colonial investments are compared in Figure 7.6, which provides a more rigorous analysis of over time trends. I model the impact of the logged distance to the nearest clinic or primary school in 1902, 1912, 1932, and 1952 on the equivalent measure in the following decades. Although there is initially a strong relationship for each in the shorter term, the effect of pre-First World War investments fades by the early postcolonial period. A similar pattern is seen in the distances to services at later points in time, whose effects fade by the era of decentralization. Looking at village-specific access, my data suggest less long-term persistence at the village level than Huillery’s (Reference Huillery2009) measurement of colonial legacies at the level of the colonial cercle.
Effect of early investments on later investments. Distance from services in (a) 1902, (b) 1912, (c) 1932, and (d) 1952
Overall, the evidence is mixed for Huillery’s argument. Any early “stickiness” of colonial investments appears to have largely faded by the postcolonial era, which may mean that these starting points matter, but that over a longer run more proximate factors hold more influence.
Colonial and Postcolonial Political Favoritism
In 1960, 2 in 1,000 rural Senegalese had their certificate d’etudes primaire (CEP), but this number was much higher at 6–8 per 1,000 in the lower Casamance (present-day Ziguinchor Region).Footnote 48 That the region of Cap-Vert (home to Dakar) and the lower Casamance were better educated was largely the legacy of the early and intense efforts of Catholic missions, which encouraged families to send their children to school.Footnote 49 Patterns in healthcare access were altogether different. The Casamance, along with Louga and Senegal Oriental, was significantly underserved in health through the early 1980s.Footnote 50
These inequities – already evident by mid-century – return us to the question of what characteristics of the colonial state should bias against the persistence of precolonial legacies. Subnational variation in access may have little at all to do with proximity or spatial diffusion, but instead be driven by the political ambitions of the French colonial state, which sought first and foremost compliance and tax revenue. To this end, French strategy toward the countryside largely revolved around the peanut economy, with the Sufi Islamic Brotherhoods acting as valuable intermediaries. Despite adopting a language of “modernizing” the link between France and her colonies in the 1950s and the introduction of local elected assemblies, the French kept its political negotiations focused around an urban political elite while granting autonomy to rural brokers, often Sufi leaders, in the countryside.Footnote 51 The postcolonial state demonstrated unusual continuity in these priorities, by and large relying on the same stratum of rural elites that had worked closely with the French.Footnote 52
This produced distinct tensions for the postcolonial state. “Political integration,” which sought to reduce inequalities in access to social services and to bring areas that had been largely isolated during the colonial era more fully into the country’s economy, figured prominently in the first four-year plan of President Senghor and his Prime Minister, Mamadou Dia.Footnote 53 The early postcolonial state’s efforts to transform the lives of the peasantry resulted in more than 50 percent of the state’s annual spending going to the “social budget” in the first years of independence.Footnote 54 At independence in 1960, Senegal counted 106,911 primary school students, a number that rose by more than 20,000 to 127,000 in 1962.Footnote 55 By and large, primary school construction and the construction of rural dispensaries were among the most popular services provided by the newly independent state.Footnote 56 This trend eventually abated in the late 1960s, however, as Senghor’s government realized that it could not support the costs associated with such rapid expansion, notably salaries and facility maintenance.Footnote 57 The government also emphasized preventative rural health, but even these early goals were never met as French medical training and practices continued to dominate. By 1974 the country had only 281 doctors.Footnote 58 Primary school construction slowed considerably as did the growth in enrollment rates, which went from a 11.2 percent annual increase in 1964 to a mere 2.7 percent in 1971.Footnote 59 Although reaching complete primary school enrollment remained a goal in subsequent government plans, the expected date of this achievement was pushed back to 2000, with short-term objectives hovering around 40 percent. Senegal has yet to meet the goal of complete primary school attendance; in 2019, it hovered around 75 percent.
Did the expansion of services by the colonial and/or postcolonial regime map onto these political objectives? If the colonial and postcolonial states systematically allied themselves in rural areas with the Wolof, peanut-producing heartland, we might expect investments to cluster in these areas. This would be the key prediction of Roessler et al. (Reference Roessler, Pengl, Marty, Titlow and van de Walle2018) who find that colonial development centered around cash crops zones, generating persistent spatial inequalities in African states to the present. Alternatively, we might expect goods to accumulate in the Mouride heartland. Figure 7.7 presents a series of OLS regressions estimating the impact of numerous potential sources of political favoritism on a village’s access to social services over time. I begin by looking for evidence of ethnic favoritism to Wolof and Serer (Figure 7.7a) dominant local governments. These ethnic groups were closest to Dakar and traditionally inhabited much of the peanut basin. Second, Figure 7.7b presents the results of models examining whether areas that produced large amounts of peanuts, Senegal’s main export crop, in either 1900 or, later, 1950, were favored. This allows me to account for both the original peanut-producing zones and the later expansion of production southeastward as the soils of Cayor and the Djoloff were exhausted. Finally, Figure 7.7c looks at the role of the Mouride Brotherhood, first, measured by the number of Mourides living in each canton in 1917 as reported by Marty (Reference Marty1917) and, second, whether the village falls within the large swaths of land opened to Mouride pioneers as estimated by Pélissier (Reference Pélissier1966, 316). Geographic controls and model specification are equivalent to those described above.
Colonial favoritism on service access over time: (a) ethnic favoritism, (b) peanut economy, and (c) Mouride Brotherhood
This is largely a story of null effects and, when these measures are statistically significant, it is often in the opposite direction than what is suggested in the literature. There is no evidence of ethnic favoritism, with the exception of 1932 when Wolof zones do appear to briefly see more improvements in service access. The peanut basin never significantly predicts new service delivery and, to the contrary, early peanut-producing zones have poorer access to social services until the 2000s. Nor is there a significant difference for a village’s falling in the heart of the peanut economy as measured by its expanse in the 1950s. French colonial investments in clinics and schools, this suggests, were not targeted to key economic zones. Of course, favoritism likely took other forms, such as the distribution of agricultural inputs or the digging of wells in addition to the ways in which these economies shaped rail and road networks, but for the social services under study here, there is no evidence that the political ambitions of the colonial or early postcolonial state are driving the patterns I document.
Finally, villages that fall in the approximated boundary of Mouride pioneer settlements only see a benefit in the early 1970s. There is, however, a positive and significant effect for villages falling in cantons with high numbers of Mourides in 1932 and 1952 and, interestingly, again in 2002. This certainly reflects the powerful role played by the Mouride Brotherhoods in the late colonial and postindependence era. This effect disappears by 2012.
Whither the Colonial State?
The data presented above further substantiates my claim that despite their magnitude, the changes induced by French colonization cannot explain the contemporary politics of public goods delivery in decentralized Senegal. This runs counter to a long-standing assumption among scholars of political economy that we can locate fundamental causes of contemporary development disparities in the colonial period. Although proximity to the colonial state heavily influenced access to social services through the 1970s, my data suggest that the effects of colonialism on local service access have faded by the early 2000s. The contemporary politics of redistribution in rural Senegal do not appear to be driven by exposure to the colonial state, to path dependencies in its social service delivery, or by the political or economic logics of colonial rule.
Certainly, colonialism had myriad influences that I do not begin to capture here. I do not question that colonialism had broader, reverberating impacts in rural West Africa nor that there are other pathways through which colonialism impacts the present.Footnote 60 Senegalese politics is still heavily oriented toward appeasing urban citizens and working via maraboutic networks in rural zones, for example. Still, my findings echo an argument that it is no longer empirically or theoretically useful to speak of a postcolonial state after the changes brought about by the post-Cold War world of structural adjustment and democratization (no matter how thin).Footnote 61 If this is true, then it is not surprising that my data document a fading role for colonial legacies in service provision. Figure 7.6 suggests that some colonial influences have washed out by as early as the 1970s. Maseland (Reference Maseland2018) similarly finds that within two decades of Nigeria’s independence, the impact of colonialism on per capita income and institutional quality had largely faded. This may not be limited to sub-Saharan Africa. Though identifying disparities in contemporary development outcomes between areas of India that were ruled directly and those ruled indirectly, Iyer (Reference Iyer2010) observes that this gap is narrowing over time as a function of the postcolonial state’s own policies. Collectively, this suggests that colonialism may be influential but not deterministic, highlighting the risk of engaging in what Cooper (Reference Cooper2005, 17) has called “leapfrogging legacies” that fail to unpack historical processes and, in the process, undermining the agency of local actors and the ambitions of postcolonial leaders and bureaucrats alike.
Discussion: Why Don’t All Legacies Persist?
This chapter has shown evidence of two legacy effects: one that reappears after over 100 years and one that fades over time. In unpacking both precolonial and colonial legacies during the course of the twentieth century, I have attempted to respond to one of the most prominent criticisms of recent work on the long-run historical causes of contemporary political and economic development: scholars make “heroic assumptions” about historical processes that strain credibility by omitting decades of history.Footnote 62 This “compression of history” relies on two assumptions about unit homogeneity, the idea that causal processes unfold the same in all cases.Footnote 63 First, this work supposes that historical institutions have universal effects and, second, that this places certain countries or communities in different equilibria that become path-dependent. The dominant approach in the “historical renaissance” therefore is that history matters because it can explain why some societies become “stuck” on more or less developmental paths.Footnote 64
This is a powerful view of history, assigning great weight to say the intergenerational persistence of attitudes (as argued by Nunn and Wantchekon Reference Nunn and Wantchekon2011, for example) or to the lock-in effects of institutions. Acemoglu et al. (Reference Acemoglu, Johnson and Robinson2001), to take a prominent example, argue that institutions brought by settlers have sunk costs, generating path dependencies as the cost of switching to other institutional forms rises for subsequent generations. Huillery (Reference Huillery2009), introduced above, makes a structurally similar argument: French colonial investments placed some districts on a “virtuous circle,” as the at times happenstance placement of a school in one district over another created long-run path dependencies in investments. This perspective thus predicts persistence, meaning that change should only arise following strong shocks, such as those that correct an unproductive equilibrium or, to embrace a different language, that produce critical junctures for institutional change. Putting aside, if we can, the question of more or less plausible historical processes, one of the most remarkable features of this scholarship is that there is limited room, if any, to consider whether and when legacies simply fade away.
Yet, this is exactly what I observe. My data on social service access find no evidence that the French strategically targeted certain areas of the country over others in their early colonial efforts, but rather settled according to convenience. Once faced with the need to actually govern vast swaths of territory, early investments were largely driven by the various attitudes of local administrators, choosing to invest in areas over which they had the most oversight, effectively meaning their own settlements and those in close proximity. When the French government was finally willing to invest in its colonies, social services expanded outward in an attempt to equalize access across the country, a project largely continued in the postcolonial period. To the extent that the colonial state had political and economic objectives, I find no evidence that this was systematically correlated with precolonial centralization or even investments made during earlier phases of the colonial state construction.Footnote 65
This suggests that not all historical legacies persist equally and, in turn, that we have insufficient leverage over why and when they are more likely to do so. One clear implication of my findings is that colonial effects may be better modeled as a slow evolutionary process than as equilibria that necessitate an exogenous shock to create change. This should not be a radical position. To the extent that scholars have established how empires (e.g. Barkey Reference Barkey2008) or the colonial state itself (e.g. Young Reference Young1994) evolve, consolidating and eroding their authority over time, it is eminently rational to expect the same from the legacies they leave behind. This makes equilibrium approaches ill-suited to explain the patterns I document in this chapter. Rather than isolating one discontinuous moment of change, development initiatives in Senegal’s countryside have adapted and responded to new political imperatives: peasant malaise led to the creation of rural communities early on, but it was structural adjustment and the political upheaval of the early 1990s that generated the move toward Acte II, for example. Past investments may be sticky, but they do not appear to be “stuck.”
Yet aren’t my findings suggestive that rural Senegal is simply returning to a longer-run equilibrium? In many ways, this is the case. As laid out in Chapter 2 and substantiated with data presented in Chapters 4 and 6, precolonial legacies are tied to the persistence of local elite social status, which validates and reifies narratives of descent from a centralized precolonial past. I theorize that politically latent social institutions are redeployed for new purposes as political actors adapt to changing institutional circumstances, such as the advent of decentralization.Footnote 66 The social institutions that I identify as critical may be an equilibrium, therefore, but this only matters for the outcomes of interest under specific formal institutional configurations.
In this vein, I suggest that contemporary differences can be traced to the distant past but that this may not become causal until an exogenous institutional shock fosters institutional congruence. Paying attention to such political interactions is critical. For example, Acemoglu, Reed et al. (Reference Acemoglu, Reed and Robinson2014) observe a strong legacy of indirect rule in Sierra Leone, but note that such persistence was not uniform. In states such as Ghana or Uganda, the presence of a large, centralized precolonial state generated chiefs with sufficient authority to challenge postindependence elites. In the face of these challengers, elites consolidated power to rule more directly as a means of co-opting potential rivals, a political strategy endogenous to local interactions between formal and informal institutions. If we recognize the redistributive effects of institutions, then any shift in the power coalitions controlling state institutions should reshape who benefits from social service investments.Footnote 67 Crucially, this can happen even absent major institutional shocks.
As Cooper (Reference Cooper2005, 25) has argued, “there is a danger that ahistorical history encourages an apolitical politics.” By unpacking history, I have attempted to restore some agency to both the postcolonial state, whose policies appear to have at least sometimes muted the disparities of the colonial era, as well as the agency of local communities to embed institutional reforms in their own local political narratives. If the appeal of legacy arguments lies in our recognition that “real social processes have distinctly temporal dimensions,” my framework offers a means to reinsert social mechanisms more clearly into our understandings of why things do (and do not) persist over time.Footnote 68
One potential reading is that institutions imposed from above, such as the extractive logics of the colonial state, may be more prone to peter out rather than to persist, while the more organic legacies of the precolonial past retain an ability to adapt to changing political, social, and economic realities as they are repurposed by local communities at the grassroots. This is because such “cognitive” legacies are most likely to be resilient within closed groups and when they invoke strong prescriptive norms.Footnote 69 Still, the social institutions at play in rural Senegal may both persist for generations and remain particularly vulnerable to shocks. Still, that they have survived the colonial encounter as well as the transition to independence suggests a remarkable durability. At base, however, the ability of social institutions to shape development outcomes is contingent on their overlap with formal institutions. This means that if decentralization was to be abolished or, more limitedly, if authority over public goods was to be transferred back to the state, then the effect I observe should disappear.
Conclusion
Together, the findings of this chapter reveal that historical legacies may face different prospects for persistence. While understanding the contours of the temporal variation of our outcomes under study is a valuable empirical pursuit in its own right, it simultaneously allows me to question the assumption behind the majority of mechanisms that have been put forth in work in the “historical renaissance”: that the geography of subnational inequalities has not dramatically changed and that these disparities have amplified over time.
In contrast, I reveal changing landscapes of political logics and political favoritism. This reinforces my theory of institutional congruence because the legacy of the precolonial past is dependent on formal institutional arrangements. It is only when formal and informal institutional boundaries are congruent that we see an effect on precolonial statehood. This results in a striking ability of precolonial geography to shape outcomes in the present, but this is not the result of a uniform or uninterrupted process. In this way, my argument can be read as a response to the recent tendency to pretermit decades, if not centuries, of history within the historical renaissance sweeping development economics and political science. In rural West Africa, historical antecedents matter, but without understanding when and for how long leaves us vulnerable to assuming improbable causes. This pushes forward debates on institutional persistence by specifying clear mechanisms of persistence, as called for by Nunn (Reference Nunn2009), while remaining sensitive to how plausible those mechanisms are at different points in time.
The second argument advanced in this chapter is that not all legacies persist with equal robustness to the present. Studying what fades away can be as intellectually useful as studying what persists. In the face of mounting evidence that important subnational trajectories exist in developing countries and that they can be explained by the distant past, it is imperative that we take seriously how these processes actually work themselves out over time. Scholars have been quick to highlight the need to disaggregate by type of public good or across countries.Footnote 70 By demonstrating the value of unpacking temporal process, this chapter suggests that we must also take seriously who makes these decisions and when.
Open access