Theory

Implanting early representative institutions required a group willing and able to lobby for such reforms. Although European settlers tended to exhibit economic and political influence wherever they settled in large numbers, many originated from metropoles that did not contain representative institutions. Only European settlers emanating from a home country with a representative tradition should promote early electoral institutions, which empirically distinguishes Britain from other major Western European colonial powers.

Regarding institutional evolution, for settler colonies that established early representative institutions, class-based theories of political transitions suggest an important impediment to maintaining representative institutions and to broadening the franchise to create full democracy. Large-scale resistance by European landed interests to perpetuate their political power carried negative implications for both democratic participation and contestation. This contrasts with Dahl’s argument that elites with a history of limited representative institutions should peacefully incorporate the masses into the polity,Footnote ¹⁸ and highlights a core democratic contradiction in European settler colonies.

Institutional Origins: Metropolitan Political Institutions

Why did some colonies gain early representative institutions? Existing research stresses the strong organizational position of European settlers that enabled them to demand political representation.Footnote ¹⁹ Considerable evidence, summarized below, indeed establishes the dominant economic and (often) political position that European settlers commanded. This contrasts with nonwhites who were subjugated in the colonial order and whom European officials thought unfit for self-rule.

Only European settlers who descended from a metropole that itself had a representative tradition should create early representative institutions. The general logic for qualifying the institutional origins hypothesis is straightforward. International powers with more liberal domestic political institutions are more likely to promote liberal institutions elsewhere, and the position of domestic elites on democracy promotion depends on their relationship with liberal polities. Much existing research on this topic focuses on post—Cold War actions by the United States and the European Union to promote democracy in the excommunist world and elsewhere. Their stances contrast with authoritarian powers, such as China and Russia, which either do not make electoral institutions a precondition for support or actively oppose democratic institutions. The beliefs and incentives of citizens within target countries should also matter. For example, Steven Levitsky and Lucan Way argue that in the post—Cold War world, countries with high “Western linkage” are likely to democratize because economic, cultural, and communication ties between their citizens and Western countries create greater desire for democratic institutions.Footnote ²⁰ Although their theory does not require elites in non-Western countries to have emigrated from the West to hold these preferences for democratic institutions, such elites are akin to European settlers in the setting discussed here because affinity toward liberal institutions causes demands for electoral reform.

Did European colonial powers exhibit important differences in their representative institutions? Across several centuries of European colonial rule, Britain (and earlier, England) exhibited a stronger representative tradition than did Spain, Portugal, or France. Figure 2 depicts constraints on the executive for these four major Western European colonizers over fifty-year intervals between 1600 and 1950. The data draw from the Polity IV data set and from Acemoglu, Simon Johnson, and Robinson.Footnote ²¹ Each data point averages twenty-year windows before and after the stated year.Footnote ²² Smoothing the data provides snapshots of the differences in metropolitan executive constraints across European empires over time without depicting sharp fluctuations in democratic constraints at various periods (for example, the struggle in England between the Crown and Parliament during most of the seventeenth century).

Figure 2

Metropolitan Executive Constraints in Half-Century Snapshots

A notable trend in Figure 2 is that Britain’s constraints on the executive grew during the first major periods of imperial expansion and contraction, which David Abernethy dates between 1415 to 1773 and 1775 to 1824, respectively.Footnote ²³ Kevin Narizny compares estates in medieval and early modern Europe and concludes, “Only in England did a medieval assembly evolve into a representative parliament with sovereign authority over the crown, and only in England was liberal protodemocracy a stable equilibrium.”Footnote ²⁴ Especially after the Glorious Revolution in 1688, Britain exhibited parliamentary constraints on the monarch unmatched by other major colonizers,Footnote ²⁵ and British settlers strongly imbued representative norms.Footnote ²⁶ By contrast, the Spanish monarchy retained absolute power until the Napoleonic Wars in the early nineteenth century, which caused it to lose most of its American colonies.Footnote ²⁷ Collectively, the British and Spanish American empires accounted for almost every colony with a sizable European population during this period.

Britain also differed from other European powers with settler colonies during the second major waves of expansion (1824 to 1912) and contraction (1940 to 1980).Footnote ²⁸ The major migration of Portuguese settlers to Angola and Mozambique starting in the 1930s began during Antonio de Oliveira Salazar’s dictatorship,Footnote ²⁹ which had the lowest possible Polity IV executive constraints score. France represents a mixed case. It exhibited high executive constraints between 1877 and 1939, and again between 1947 and 1957. But unlike Britain, France exhibited prolonged struggles between authoritarian and democratic forces throughout the nineteenth century, in the twentieth century during World War II, and again the late 1950s with the establishment of the Fifth Republic. Even during democratic periods, France’s politics were unstable compared to Britain’s.Footnote ³⁰ For example, elected officials in France’s Fourth Republic were susceptible to pressure from special interests, such as European settlers and the military, due to unstable governments and weak party discipline.Footnote ³¹ Furthermore, Britain and France practiced different colonial governing philosophies. Although some scholars exaggerate the differences between Britain’s indirect rule policies and France’s preferences for more centralized control, variance in delegation practices did meaningfully affect prospects for institutional transplantation.Footnote ³² For example, France “tightly controlled” European settlement in French Algeria and “the Algerian enterprise received much greater governmental supervision and the population was subject to a greater degree of regulation, unthinkable in a contemporary British colony.”Footnote ³³

Overall, these differences, which spanned centuries, implied that colonial officials and settlers in the British Empire had a stronger representative tradition on which to draw, yielding:

This hypothesis relates to broader debates about the importance of colonizer identity and metropolitan institutions. Research on European settlers usually de-emphasizes the importance of colonizer identity,Footnote ³⁴ which echoes broader shifts in the colonialism-democracy literature.Footnote ³⁵

These accounts instead argue that selection effects explain any British colonial distinction. For example, Acemoglu, Johnson, and Robinson claim, “It appears that British colonies are found to perform substantially better in other studies in large part because Britain colonized places where [large-scale European] settlements were possible, and this made British colonies inherit better institutions.”Footnote ³⁶ Hariri argues that British and Spanish settlers drew from similar legacies because neither metropole was fully democratic in the eighteenth century,Footnote ³⁷ and that Spanish-American settlers created “a system of comprehensive checks and balances” during the colonial era that “facilitated the spread of early representative institutions.”Footnote ³⁸ Similarly, Robert Woodberry argues, “Some scholars suggest that British colonialism fostered democracy . . . but this may be because [Protestant missionaries] had greater influence in British colonies.”Footnote ³⁹

Although several studies on economic development argue that the beneficial effects of European settlers are limited to British colonies,Footnote ⁴⁰ my argument emphasizes distinct considerations—focused on political institutions—about British colonialism. Matthew Lange, James Mahoney, and Matthias vom Hau expound the distinction between British liberal economic institutions and Spanish mercantilist institutions,Footnote ⁴¹ and Mahoney compares differences in mercantile and liberal Spanish economic institutions over time. But the more theoretically relevant focus for studying democracy concerns the differences in Britain and Spain’s political institutions.Footnote ⁴² Matthew Fails and Jonathan Krieckhaus appeal to a broad range of factors that distinguish British settlers and argue that British settlement is essentially a binary variable that differentiates only the neo-Britains from the remainder of the empire.Footnote ⁴³ But Britain also colonized numerous territories in the Caribbean that, despite featuring smaller European populations, nonetheless drew from a similar representative tradition as contemporaneous North American settlers.Footnote ⁴⁴ Hypothesis 1 also applies to these colonies.

Institutional Evolution: Landed Oligarchs and Resistance to Franchise Expansion

Did early representative gains yield a smooth path to eventual democratization? Applying mechanisms from class-based theories of democratization and democratic consolidation to settler colonies demonstrates the strong incentives of landed European elites to block expansion of political participation. Landed elites feature centrally in class-based theories, which have a long history in political science. Barrington Moore famously proposed “no bourgeoisie, no democracy,” which recent research expounds.Footnote ⁴⁵ Other studies focus on either the working classFootnote ⁴⁶ or the interplay between the working class and political elites.Footnote ⁴⁷ Regardless of the specific actor posited to promote democracy, class-based theories agree that landowning elites should repressively resist franchise expansion, especially in circumstances of high land inequality. Boix and Acemoglu and Robinson posit one plausible mechanism for this.Footnote ⁴⁸ Their theories examine how a minority elite group and the masses strategically interact. The masses may pose a revolutionary threat because of their numerical superiority, which enables them to achieve concessions from the political/economic elite. But elites who control political power amid high economic inequality face incentives to repress rather than to expand the franchise to include the masses, who would redistribute considerable income from the elites to themselves. Landlords particularly fear majority rule because land is a nonmobile asset that is relatively easy to redistribute.Footnote ⁴⁹

Existing theories focus on the incentives of landowning elites to prevent franchise expansion, and examine democratic participation rather than contestation. But two mechanisms can account for why actions to resist broad political participation should also diminish the quality of contestation. First, it may be possible for elites to maintain low participation—at the expense of sacrificing high contestation—by delegating authority to an authoritarian strongman who can better counteract threats from below. For example, Dan Slater argues that serious threats from below cause elites to replace democratic representation with authoritarian “protection pact” institutions that can better counteract the threat, as in postcolonial Malaysia and Singapore.Footnote ⁵⁰ Second, antienfranchisement repression should foster more extreme opposition leaders. Mehdi Shadmehr shows that higher repression deters moderates from participating in organized antigovernment movements because moderates are less willing to pay the associated participation costs.Footnote ⁵¹ Relatedly, repression should also raise the likelihood of fostering extremist opposition leaders who have a comparative advantage in coercion rather than in electoral participation,Footnote ⁵² perhaps in the form of revolutionary vanguard parties.

The conditions under which class-based theories expect democratic resistance to expanded electoral participation closely match empirical conditions in colonies with a sizable European minority, most of which exhibited highly unequal land distribution patterns between Europeans and non-Europeans, as discussed below. Although the European settlers controlled assets in addition to land, many of these colonies were founded by displacing natives or by settling forced migrants onto European-controlled plantations, thereby making land a crucial source of economic and political power. In earlier work, I discuss how European land control in African settler colonies created broad interests against majority rule even among nonfarming whites.Footnote ⁵³ Through land and other assets, colonial European settlers wielded considerable political influence, and by lobbying the metropole or by directly controlling the top political offices they could achieve their preferred economic policies, such as controlling the best land or distorting the labor market.

There are two additional scope conditions for when European settlers should engage in widespread resistance to electoral participation by nonwhites. First, resistance should occur only if European settlers are politically powerful—although this power could diminish over time, as occurred in the British Caribbean. Second, heavy repression should not be necessary to exclude nonwhites from political participation in territories with large European majorities that lack a sizable threat from below. Therefore, the overall relationship between European population share and incentives to exercise heavy repression should be nonmonotonic. Most constituent territories within the neo-Britains exhibited large European majorities. Although the native populations were smaller and the land was less densely populated in North America and Oceania than in many other parts of the world, aggressive European expansion early in the colonial period and natives’ lack of immunity against European diseases, which resulted in genocide-magnitude population decline, enabled the preponderance of Europeans. The one exception among the neo-Britains is the colonies/states in the American South. Because millions of enslaved Africans were forcibly moved to the region, it largely fits the scope conditions under which elites should exhibit large-scale resistance toward franchise expansion. Overall, these considerations imply:

Although strategies to defend elite privilege are central to class-based theories, existing colonialism research mentions this mechanism only in passing. Fails and Krieckhaus argue that British colonies other than the neo-Britains did not exhibit meaningful variation in settler population size, and therefore medium-size British settler colonies should not differ from colonies largely devoid of European settlement. They also briefly mention that by creating an interest group that favored extractive economic institutions, small Spanish settlements could have engendered worse development legacies than colonies without settlement, which resembles my argument.Footnote ⁵⁴ Lange, Mahoney, and vom Hau distinguish British from Spanish colonies based on liberal/mercantile institutions, but differences in economic institutions are less important for explaining democratic trajectories.Footnote ⁵⁵ Mercantile policies might contribute to creating “entrenched actors who benefit from state privileges,”Footnote ⁵⁶ but many British colonies in the Caribbean and in southern Africa contained similarly privileged European elites—even though Imperial Britain and Spain pursued divergent economic policies.

Hypothesis 2 is also theoretically intriguing because juxtaposed to Hypothesis 1 it shows how an explanatory factor can yield divergent implications for different components of democracy, a largely novel consideration for colonialism research. Much research on European settlement focuses on the distinct outcome of economic development, and studies specifically on democracy tend not to disaggregate its components. Although Acemoglu and colleagues and Hariri discuss one positive effect of European settlement on democratic contestation,Footnote ⁵⁷ neither they nor their critics scrutinize how entrenched settler oligarchies’ incentives to restrict franchise expansion can undermine earlier contestation gains.

Assessing Institutional Origins

Did most varieties of European settler colonies experience early representative institutions, or was this feature largely limited to British colonies? Analyzing newly compiled data on elected colonial legislatures from the seventeenth to twentieth centuries supports Hypothesis 1. Statistically, British settler colonies—but not settler colonies outside the British Empire—are associated with elected legislatures before 1945.

Data

This section briefly describes the data in Figure 3 and Table 1 (both below). Section A.1 of the supplementary material provides more detail.Footnote ⁵⁸ Table A.1 of the supplementary material lists every territory in the sample, the first year with a colonial legislature, score on the settlers’ variable, and colonizer. Table A.2 provides summary statistics.

Sample

The sample consists of a panel of 144 former Western European colonies and begins in 1600.Footnote ⁵⁹ It includes numerous small islands in the Caribbean and Pacific, including several present-day dependencies. Due to data constraints, in most cases the units correspond to modern-day countries, with exceptions for Spanish American states in which the postcolonial countries do not correspond with colonial units, and six ex-British countries that combined multiple colonies at independence or after a lengthy existence as distinct colonies (six in Australia, four in Canada, four in South Africa, two in St. Kitts and Nevis, two in Trinidad and Tobago, and thirteen in the United States). Temporally, the sample includes only years under colonial rule.

Colony-Wide Elected Representative Body

This article introduces self-collected data for each colony on elections to a colony-wide representative body. A coding document provides extensive details and the sources used to code this variable.Footnote ⁶⁰ The nature of elected representative bodies differed widely across empires and over time within empires. In some cases, such as assemblies in seventeenth-century British America, these elected bodies possessed extensive legislative powers. In other cases, such as many legislative councils in the British Empire from the nineteenth century on, these bodies were considerably less powerful than the colonial executive was and at least one, but not all, members were elected. For British colonial legislative councils and similar bodies in other empires, the coding requirement is that at least one member was elected rather than all members or even a majority. For example, although Britain created a legislative council for St. Lucia in the 1830s, the first year with any elected members is 1924, the year the data set uses. In other empires, representative bodies like the delégations financières introduced in Algeria in 1898 or the Volksraad introduced in Indonesia in 1916 lacked formal legislative powers and were purely advisory but meet the criteria of an elected representative body.Footnote ⁶¹ In addition, the colony-wide criterion excludes local bodies, such as town councils (cabildos; see below for more detail) in Spanish America and municipal councils. It also excludes elections to an empire-wide legislature, which France introduced in 1789 and allowed intermittently throughout the nineteenth century, because these legislatures did not grant colonial citizens or subjects voice over their own governance. In most colonies prior to World War II, the population percentage that could vote (if any) was very small.

The new data set documents colonial electoral bodies across a broader sample and period and provides more extensive documentation than earlier data sets. Most standard democracy data sets provide only post-independence data. Varieties of Democracy (v-dem) provides data on franchise size in colonies in the twentieth century,Footnote ⁶² but lacks earlier information or coverage for many smaller countries. A recent expansion of v-dem extends back to 1789, but Historical v-dem only covers countries that gained independence before 1900 and thereby excludes the bulk of the Western European colonial world. Fifty-seven colonies in my data set gained electoral representation prior to 1900, including thirty before 1789. The Political Institutions and Political Events data set also provides some information on legislative elections, but only exhibits widespread coverage of years under colonial rule after 1945.Footnote ⁶³

European Settlers Variable

The main European settlers variable indicates whether a territory contained a European population share of at least 5 percent at any point in the colony’s history.Footnote ⁶⁴ Several considerations motivate using this simple binary measure: the panel spans a very long period, some countries fluctuate considerably in European population share over time, and data on colonial European populations are inherently uncertain further back in time. Although 5 percent may appear to be a low threshold, the cases discussed below show that even colonies with relatively small European minorities fit the scope conditions of the theory that the European settler population is large enough to exhibit political influence. To show that the results do not depend on a particular population threshold, robustness checks analyze a logged continuous European population—share variable that fluctuates throughout the colony’s history.

Colonizer Identity

Colonizer identity is based on the final Western European country that ruled the territory. The sample excludes all years prior to the final colonizer gaining control (see Section A.1 in the supplementary material).Footnote ⁶⁵ For example, because Britain conquered Mauritius during the Napoleonic Wars, the sample includes Mauritius as a British colony from 1814 until independence but excludes it before 1814.

Main Patterns

Figure 3 shows the percentage of colonies with an elected representative body between 1600 and 1959, disaggregated by settler/non and British/non. Panel (a) codes a colony year as 1 if the colony ever had an elected colonial representative body, and 0 otherwise. Because the dependent variable is whether a territory has ever had representative institutions, percentage dips occur either because a new territory entered the sample and did not immediately gain elected representation or because a colony with a representative body gained independence.Footnote ⁶⁶ The cutoff year for panel (a) is 1959. The percentages are exceedingly difficult to interpret after 1959 because the number of colonies dropped precipitously in the 1960s, generating rapid fluctuation in the sample. Panel (b) shows how the sample changes over time by presenting the number of colonies by category through 2000.

Figure 3

Elected Colonial Representative Bodies Since the Seventeenth Century

Analyzing three distinct periods in Figure 3 yields the main takeaways.

Early Colonies

Until the mid-nineteenth century, elected representative bodies were exclusive to British settler colonies. All colonies founded by English settlers in North America and the Caribbean and some colonies founded by British conquest created elected legislatures shortly after colonization. Starting in the 1840s, similar political developments occurred in British colonies in Oceania and in southern Africa. Jack Greene discusses New World colonies and shows evidence that for Englishmen, liberty was “not just a condition enforced by law, but the very essence of their national identity.”Footnote ⁶⁷ Settlers’ colonial assemblies consciously sought to replicate the English House of Commons and to obtain corresponding political privileges.Footnote ⁶⁸ British North American colonies largely controlled their internal affairs, and their legislatures outpaced the House of Commons in terms of autonomy due to their “continuous and continuing British connection and the tremendous impact of the British constitution upon their own perception of the constitutional order.”Footnote ⁶⁹ Even in small Caribbean islands with less ability to resist metropolitan British encroachment, legislatures exerted considerable autonomy, fully controlling finances and exercising extensive executive powers.Footnote ⁷⁰

These British institutions contrast sharply with the despotisms of eighteenth-century Spanish, Portuguese, and French American empires.Footnote ⁷¹ Samuel Finer quotes Adam Smith’s The Wealth of Nations, written in 1776: “In everything except their foreign trade, the liberty of the English colonists to manage their own affairs in their own way is complete. . . . The absolute governments of Spain, Portugal, and France, on the contrary, take place in their colonies.”Footnote ⁷²

Spain, which possessed most of the remaining American colonies at the time, practiced direct authoritarian rule. The Spanish Crown did not legally permit colonial officials to perform any executive or legislative functions. “Formal power was not shared by anyone outside the immediate Council and the king,”Footnote ⁷³ local officials functioned solely as judiciaries, and no colony-wide parliamentary bodies were established.Footnote ⁷⁴ Cabildos were the one institution with some popular participation at the local level, but shortly after towns were formed the Crown typically diminished the power of cabildos and sold the office to raise revenues.Footnote ⁷⁵ “As a repository of people’s liberty, a training school for the democratic system to be set up after independence, the cabildo possessed no potency at all. It had little or no freedom in action or responsibility in government. Its weakness was not a recent development at the turn of the nineteenth century. On the contrary, the institution had been in a state of collapse for generations.”Footnote ⁷⁶ The first and only attempt to promote general elections for the Spanish American Empire occurred in 1809 in response to turmoil in Spain caused by the Napoleonic Wars, but even these elections were to an empire-wide assembly in Spain rather than to local legislatures—and colonial representatives were never seated in the Junta Central.Footnote ⁷⁷

These differences also highlight the importance of colonizer identity relative to natural endowments.Footnote ⁷⁸ At the turn of the nineteenth century, elected legislatures pervaded British territories regardless of whether the territory was suitable for small-scale farming (former colonies in the northern United States and Canada) or for sugar plantations (much of the Caribbean), and the exceptions arose because of shifts in British colonial policy (discussed below). Spain imposed authoritarian institutions across South America, Central America, and the Caribbean despite varying endowments, as did France among its Caribbean sugar colonies and in Quebec prior to 1763.Footnote ⁷⁹

Late Nineteenth Century

Several settler colonies outside the British Empire gained electoral representation starting in the mid-nineteenth century. Shortly after the 1848 Revolution in France and the establishment of the Second Republic, Guadeloupe and Martinique in the Caribbean and nonsettler Réunion in the Indian Ocean each created a conseil général; French Guyana and nonsettler Senegal followed suit several decades later.Footnote ⁸⁰ Whites in Algeria (France’s primary settler colony) gained representation at the end of the nineteenth century, but French Morocco never gained a legislature, and neither did the authoritarian-ruled Portuguese settler colonies in Africa prior to 1945. Rupert Emerson qualifies the importance of representative assemblies in centrally ruled French colonies: “Despite the revolutionary tradition of liberty and equality, the French colonies offered little in the way of democratic institutions . . . At best the French created advisory councils of a dubiously representative kind with some financial and administrative powers but little general legislative competence,” consistent with France’s strong propensity for direct colonial rule.Footnote ⁸¹

According to Stephen Roberts, during the last decades of the nineteenth century amid debates about administrative issues in Algeria, “There was no question of self-government at all—no thought that the French colonies should follow the English in going from oligarchic to representative and then to responsible government.”Footnote ⁸² Even electoral reforms implemented in 1898 “did not envisage anything in the nature of the English autonomy or self-government: it simply meant the development by French officials as before, but in the new direction of the colony’s own interests,” as opposed to the earlier policy of controlling Algeria “from Paris and on exclusively French models.”Footnote ⁸³

Post–World War I

By the 1930s, many nonsettler colonies had established elected legislatures, including British India (1910), British Nigeria (1923), and French Mali (1925). Only in the decades after World War II did British nonsettler colonies and non-British colonies converge to British settler colonies. France introduced legislative elections across its sub-Saharan African colonies in the 1940s and 1950s, Britain gradually decolonized its entire empire, and Portugal belatedly attempted to gain African support of the colonial project in the early 1970s by introducing elections.Footnote ⁸⁴

Statistical Evidence: British Settler Colonies and Early Elected Representation

Table 1 statistically assesses correlates of early electoral representation under colonial rule. It uses the same data as in Figure 3, but the sample ends in 1945, the beginning of the terminal colonial period, to correspond with early elected representation. The table presents estimates from a series of logit models with standard errors clustered by colony. The dependent variable captures election onset, equaling 0 in all years under colonial rule before the first year with any elected representatives and 1 in the first election year, and is set to missing in all subsequent years. Every specification contains cubic polynomials that count years since colonial rule began, and a fixed effect for early colonial onset (pre-1850). Abernethy, Ola Olsson, and others argue that the nature of colonial rule changed over time, including most empires shifting from mercantile- to imperial-based colonial rule during the midnineteenth century.Footnote ⁸⁵

Column 1 uses the binary settlers indicator for whether the colony ever had a European population share of at least 5 percent, and interacts it with British colonialism. Column 2 controls for four alternative explanations from the literature: population density in 1500, a territory’s history of statehood in 1500, a forced-settlement colony indicator, and colonial Protestant missionary population. Acemoglu, Johnson, and Robinson argue that Europeans faced difficulties creating large settlements in territories with high population density,Footnote ⁸⁶ and Hariri argues that territories with a long history of statehood were better able to resist European encroachment.Footnote ⁸⁷ Relatedly, Mahoney evaluates the effect of complexity of the indigenous society on varieties of colonialism within the Spanish Empire.Footnote ⁸⁸ Olukunle Owolabi and Woodberry each focus on other colonial-era actors argued to promote democracy, forced settlers and Protestant missionaries, respectively.Footnote ⁸⁹ Owolabi’s indicator for a large forced-settlement population is nearly synonymous with plantation colonies, and therefore operationalizes Stanley Engerman and Kenneth Sokoloff’s argument that land endowments favorable for plantation-type agriculture generated large slave populations and high inequality.Footnote ⁹⁰ As noted above, many of these scholars explicitly argue against any Britain effect, and therefore controlling for these factors addresses the most important confounding concerns identified in the literature.Footnote ⁹¹ Column 3 replaces the British colonial rule indicator with an indicator for high metropolitan constraints on the executive,Footnote ⁹² and column 4 adds the four covariates. Column 5 replaces the European settlers indicator with the continuous measure of European population share, and column 6 adds covariates.

Table 1

Correlates of Elected Representative Bodies: Colonial Rule 1600–1945^a

***p < 0.01, ** p < 0.05, * p < 0.1

^a Table 1 summarizes a series of logit regressions by presenting coefficient estimates and colony-clustered robust standard error estimates in parentheses using two-sided hypothesis tests. The bottom part of the table presents the marginal effect estimates and corresponding standard error estimates for the European settlers variables under various values of the conditioning variables.

Table 1 robustly supports Hypothesis 1. In all columns, the marginal effect estimate for European settlers is positive and statistically significant among British colonies or colonies whose metropole had high executive constraints, but it is not so among non-British colonies or colonies with low metropolitan executive constraints. In column 1, the predicted failure rate is thirty-four times higher for British settler colonies than for non-British settler colonies, fifteen times higher than for British nonsettler colonies, and nineteen times higher than for non-British nonsettler colonies. The supplementary material shows qualitatively similar results when altering the original models to end the sample in 1918 as an alternative conceptualization of early electoral representation (Table A.3), or when excluding the twenty-four colonial units within the four neo-Britains (Table A.4).Footnote ⁹³

It is also possible, in principle, that historical population density or state antiquity can explain variation in adopting elections among British settler colonies, given Mahoney’s argument that the complexity of the precolonial society affected the mode of colonial rule and John Ger-ring and colleagues’ evidence that Britain tended to more directly rule territories with a shorter history as a centralized state.Footnote ⁹⁴ However, in contrast to the Spanish American cases that Mahoney analyzes and to the sample of twentieth-century British colonies that Gerring and associates examine (which excludes many of the British settler colonies in my sample), the British settler colonies in Table 1 vary little in pre-colonial complexity. All the British settler colonies except Belize lacked any history of statehood above the local level in 1500, and all but Belize had no greater than the median level of population density in the sample in 1500.

Section A.3 of the supplementary material disaggregates British settler colonies by whether British settlement or conquest founded the colony.Footnote ⁹⁵ Whereas legal precedents enabled British inhabitants of settled colonies all the political rights of British subjects, London exercised discretion regarding whether to extend rights to conquered colonies. Furthermore, although by definition settled colonies consisted of British settlers, many conquered colonies contained sizable non-British European populations upon British conquest. Table A.5 in the supplementary material shows that British settled colonies indeed gained electoral representation earlier than British conquest colonies with sizable settler populations, although both are statistically significantly different from British nonsettler colonies.Footnote ⁹⁶

Assessing Institutional Evolution

Although only British colonies empirically support arguments that colonial European settlers facilitated early representative institutions, this finding does not rule out the possibility that European settlers across the vast British Empire (or other colonies with early representative institutions) regularly bequeathed democratic institutions that postcolonial states inherited at independence. Countering this possibility, this section analyzes institutional evolution in settler colonies and demonstrates that politically influential landed classes typically resisted franchise expansion to a rising nonwhite majority, consistent with Hypothesis 2. I analyze three regions that collectively contain most colonies with relatively early representative institutions and sizable European minorities (Africa and the British Caribbean) or tenuous majorities (the American South); see Table 2.Footnote ⁹⁷ I analyze separate eras for each region to concentrate on pivotal periods during which a previously dominant white oligarchy faced a challenge from nonwhites. For each region, quantitative evidence of aggregate patterns coupled with qualitative evidence of mechanisms supports the theoretical expectation of resistance to franchise expansion by vulnerable settler oligarchies. Section B.2 in the supplementary material analyzes informative null cases: British Caribbean colonies after World War I.Footnote ⁹⁸

Table 2

European Population Percent in British (and Select Other) Settler Colonies^a

^a Using the same settler colony threshold as in Table 1 (European population share of at least 5 percent at any point in the colony’s history), Table 2 lists every British settler colony in the New World and Africa, and all non-British settler colonies in Africa, which collectively comprise nearly every colony with elected representation at any point before World War I. Whereas the sample in Table 1 only includes the thirteen US colonies, Table 2 lists every US state that joined the Union before 1850, which provides the sample of 30 US states used in Figure 6 (see below). Table 2 lists each territory’s highest European population share percent between 1850 and whichever was later, 1900 or independence. In many Caribbean colonies, European population percent was considerably lower in the nineteenth century compared to their first decade of colonial rule in the seventeenth or eighteenth century (before mass forced migration of enslaved Africans). Section A.1 in the supplementary material describes the data (Reference PainePaine 2019c). For US states in the middle category, the second number in parentheses is the percentage of the white population residing in majority black (black-belt) counties in 1940 (Reference KeyKey 1949, 7).

Post–World War II Africa

Main Pattern: Smaller Legal Franchises in Settler Colonies

Figure 4 summarizes three distinct periods of suffrage expansion across Africa during the twentieth century, highlighting a middle period in which nonsettler colonies diverged from the settler colonies. This middle period is consistent with the theoretical expectation that when faced with a threat from below, colonies with sizable settler minorities should prevent the majority from gaining the franchise.

The sample consists of forty-three mainland African countries, including those in North Africa and Madagascar, that experienced European rule in the twentieth century and gained African majority rule after 1945.Footnote ⁹⁹ It contains annual observations between 1900 and 2000, including years before and after independence. Examining pre- and postindependence periods is useful because the timing of independence is endogenous to European settlers’ political influence, which often enabled them to delay reform. The dependent variable is the percentage of the population with the legal right to vote in national elections, measured by v-dem.Footnote ¹⁰⁰ This variable relates to legal franchise restrictions based on race, but even territories with high values of this variable are not necessarily democratic because they may lack free and fair elections. In the supplementary material, Tables A.6 and A.7 provide supporting regression analysis and Table A.8 provides summary statistics.Footnote ¹⁰¹

In the first period, the decades preceding World War II, Europeans pacified their African territories and established colonial rule. All territories exhibited a low percentage of the population that could legally vote. In fact, settler colonies exhibited larger franchises because they experienced legislative elections earlier. Europeans elected representatives in Cape and Natal in South Africa, Southern Rhodesia (Zimbabwe), and Algeria by the turn of the twentieth century, and in Tunisia and South-West Africa (Namibia) shortly after World War I. Although these franchises were restricted almost exclusively to whites, most nonsettler colonies lacked any electoral representations before World War II.

Figure 4

Legalized Suffrage in Twentieth-Century Africa (Pre- and Postindependence)

In the second period, important developments during and after World War II created changes that yielded peaceful transitions to majority rule and independence in most of nonsettler Africa,Footnote ¹⁰² but Africa’s settler colonies exhibited a divergent path from the rest of the continent. Although settler and nonsettler territories alike experienced increases in legalized suffrage in the decades following the war, this process occurred more slowly in settler colonies. The gray shaded area of Figure 4 highlights the 1955-to-1970 period and shows that nonsettler colonies expanded the franchise more rapidly than did the settler territories, as decolonization prevailed in Britain and France’s nonsettler colonies. In fact, in South Africa, the province of Cape went in reverse. Cape initiated nonracial franchise rules in the 1850s that eroded over time to a de facto ban on nonwhite political participation, which the national legislature enacted into law in 1959.

In the third period, the settler territories eventually caught up. The end of liberation wars in French North Africa (1962), Portuguese Africa (1975), and southern Africa (1980 in Zimbabwe, 1990 in Namibia, 1994 in South Africa) enabled Africans or Arabs to gain majority rule.

Evidence of white Resistance to Franchise Expansion

Considerable evidence supports the key redistributive mechanism for Hypothesis 2 posited by class-based theories: the settler landed elite repressed the majority to perpetuate settler dominance over the best land. Research by area specialists and historians of Africa supports the assertion that land inequality between Europeans and Africans was starkly higher in settler than nonsettler colonies. “In many African colonies without settlers, the colonial authorities did not attempt to disrupt local tenure practices. Indirect rule was interpreted to call for, in some places, vesting local authorities with control over land.”Footnote ¹⁰³ By contrast, almost every colony that experienced disruption to existing land-tenure practices “saw exceptionally large amounts of land alienated during white rule for the benefit of white settlers.”Footnote ¹⁰⁴ Table 3 summarizes starkly unequal land distribution patterns in four major settler colonies compared with 0 percent European land alienation in most colonies.Footnote ¹⁰⁵

European settlers did not face major challenges to their political hegemony before 1945, but post—World War II changes facilitated African mobilization and created a threat from below.Footnote ¹⁰⁶ The key economic difference between settler and nonsettler colonies—considerable European alienation of land—created broad interests against decolonization in settler colonies. For farmers, the relatively low technological barriers to farming would make it easy to replace Europeans with Africans.Footnote ¹⁰⁷ European control of the land also created positive spillovers for nonagricultural whites via broader extractive mechanisms. The major settler colonies were founded upon preferential European access to land, and Europeans also benefited because displacing Africans from their land created a cheap and mobile labor supply.Footnote ¹⁰⁸ Consequently, politically influential European settlers responded with repression rather than concessions to the African majority. Whites in South Africa and Southern Rhodesia elected extremist parties after World War II to combat rising indigenous demands, and French settlers in Algeria rigged the 1948 legislative elections to prevent Arab representation. Overall, each of the six African colonies coded as settler colonies in Figure 3 experienced a major liberation war to gain independence or, in the case of South Africa, to end European political dominance and gain majority rule amid repression intended to prevent enfranchising Africans. This is in contrast to the remainder of the continent, in which franchise expansion and independence occurred mostly peacefully.

Table 3

European Settler Land Domination in Africa^a

Source: Land data from Lützelschwab 2013, Tables 5.1 and 5.2.

^a Percentages for Algeria exclude the Sahara.

Post-Slavery British Caribbean

Main Pattern: Reversals in Electoral Representation

Figure 5 demonstrates three distinct periods of electoral representation in the British Caribbean between 1600 and 1950, highlighting a middle period in which most colonies ended elections. The middle period is consistent with the theoretical expectation that when sizable settler minorities faced a threat from below, their resistance to franchise expansion should hinder representative institutions.

Unlike for Africa, for British Caribbean colonies there is no within-region control group because they contained similarly sized European settler populations (although below I briefly discuss several colonies that retained uninterrupted electoral representation). Figure 5 uses the elected-legislature data from Figure 3, but it differs from that figure in three ways. First, Figure 5 contains only British Caribbean colonies.Footnote ¹⁰⁹ Second, it lists the number rather than the percentage of colonies with an elected legislature. Third, the legislature variable equals 1 if the colony has an elected legislature in a particular year and 0 otherwise, as opposed to coding whether the colony has ever had an elected legislature.

Figure 5

British Caribbean Colonies with Elected Legislatures

The first period entailed British settlers creating elected assemblies shortly after colonial inception, shown in Figure 5 by the close relationship between the solid and dashed lines prior to 1800. Several colonies captured during the French Revolutionary and Napoleonic wars failed to gain electoral representation (see Section A.3 of the supplementary material).Footnote ¹¹⁰ Later, a wave of electoral reversals occurred, starting with Jamaica in 1865, highlighted by the gray shaded area. By 1880, only Antigua, the Bahamas, Barbados, Bermuda, and Dominica retained any elected members in their legislatures and by 1898, Antigua and Dominica had transitioned to fully nominated legislative councils. Finally, starting in the 1920s, electoral representation again pervaded the region (see Section B.2 in the supplementary material).Footnote ¹¹¹

Evidence of white Resistance to Franchise Expansion

Why did electoral reversals occur in the second half of the nineteenth century and annul earlier contestation gains? Historical evidence closely matches the expectations of class-based theories, supporting Hypothesis 2. Most British Caribbean colonies produced sugar and by the nineteenth century featured a minority elite of European planters who ruled a vastly larger slave population. Around 1830, among nine British sugar colonies with disaggregated population data, the slave population ranged from six times the size of the white population in Barbados to more than thirty times it in Grenada.Footnote ¹¹² Sugar was either the principal or the only product in most British Caribbean colonies and plantations provided the core social and economic unitsFootnote ¹¹³ —both indicators of extreme land inequality.

In the nineteenth century, British settlers faced two types of challenges to their political power, which they exercised through elected legislatures in most colonies. First, the latent threat of revolution by the slave majority became more acute. In addition to the successful Haitian Revolution, “[s]lave rebellions significantly increased after 1815 on all the British islands. Slaves rebelled both in the major sugar colonies and on the smaller islands.”Footnote ¹¹⁴ The second challenge arose in 1833 when decades of successful lobbying by white Caribbean planters finally failed and Britain outlawed slavery throughout its empire.Footnote ¹¹⁵ Although this development created the possibility of former slaves gaining political representation, the European settlers reacted by increasing property rights restrictions on voting and creating exceptions for whites eligible to vote under the old rules.Footnote ¹¹⁶ Table 4 summarizes available voter data in several colonies and shows that less than 2 percent of the population could vote in the 1850s, even though slavery had ended more than a decade earlier. Overall, British settlers “had no intention of sharing their liberty with former slaves or of making island liberty less exclusive.”Footnote ¹¹⁷

Apprehensive of mass enfranchisement by either peaceful or revolutionary means, settlers in most colonies ultimately forfeited electoral representation and acquiesced to direct rule by the British Crown. With the end of slavery, plantation agriculture in the Caribbean became less profitable, which in turn decreased government revenues. Over time, an increasing share of white planters believed that an authoritarian government with a strong executive would increase private investment in the islands and prevent nonwhites from gaining political power.Footnote ¹¹⁸ In 1852, Britain’s Secretary of State for the Colonies warned that absent reforms, “They must anticipate being overwhelmed in the Assembly by representatives of the coloured and black population.”Footnote ¹¹⁹ The triggering event for moving to direct British rule occurred after a major revolt led by former slaves at Morant Bay in Jamaica in 1865. Although the government successfully repressed the rebellion, “the gravity of the crisis was vastly greater than anything experienced in Jamaica since emancipation.”Footnote ¹²⁰ Whites interpreted the revolt in starkly racial terms. In a speech that preceded a vote to disband the legislature, Jamaica’s governor “declared that only a strong-minded government could preserve the island from further violence.”Footnote ¹²¹ Facing largely similar circumstances, most other British Caribbean colonies followed this trajectory in the 1860s and 1870s, although in some cases the process of transforming an elected council and assembly into a single nominated council was more gradual than Jamaica’s course. First, perhaps, the council and assembly would be merged into one body, as they had been in Dominica in 1863; then the number of elected members would be reduced to leave a nominated majority; finally, the elected members would be dispensed with, and the whole legislature would consist of nominated officials.Footnote ¹²²

Table 4

Population Share of Eligible Voters in Mid-Nineteen th Century

Sources: Reference RogozińskiRogoziński 2000, 194, provides data on number of voters; Barbados population measured in 1851 and Jamaica in 1861 from Reference RogozińskiRogoziński 2000, 188; Grenada in 1829 and St. Vincent in 1825 from Reference RogozińskiRogoziński 2000, 120; and Tobago in 1775 from Reference WellsWells 1975, 253.

The three British Caribbean colonies that never eliminated electoral representation faced less dire circumstances than had Jamaica and most other sugar colonies. Neither Bermuda nor the Bahamas contained sugar plantationsFootnote ¹²³ and Barbados was “the sugar colony in which the prosperity of the planters was not imperilled and their political domination not challenged.”Footnote ¹²⁴ Although the small number of cases that do not match the scope conditions of class-based theories disallows performing statistical analysis, it is notable that the exceptions to the pattern of eliminating electoral representation correspond with theoretical expectations.

By fundamentally altering their representative system, most Caribbean settlers responded to mass threats more bluntly than other settler colonies. In contrast to the British settlers in North America, and later to those in South Africa and Rhodesia, the small size of the white plan-tocracy created severe vulnerabilities for British Caribbean settlers,Footnote ¹²⁵ yielding a preference to address their fears through direct British rule rather than independence. The possibility of moving to British Crown rule was also historically contingent. For example, settler populations in Tanganyika/Tanzania, northern Rhodesia/Zambia, and Kenya were influential, but not large enough to follow the South African or Rhodesian path of independent rule. However, after World War II, Britain developed a firm commitment to promoting electoral representation inclusive of non-Europeans.

Despite nineteenth-century electoral reversals, most British Caribbean colonies gained independence peacefully in the twentieth century and consolidated democracy after independence. Section B.2 in the supplementary material provides additional discussion of the British Caribbean in the twentieth century, showing that the nonwhite professional and working classes propelled reforms that recreated electoral representation and, later, full suffrage.Footnote ¹²⁶ Reform was possible because the political and economic influence of the white planter elite had weakened considerably by the twentieth century.

Main Pattern: voting Restrictions after Reconstruction

Figure 6 summarizes voter restrictions between 1850 and 1975 among US states. It highlights a middle period in which the South diverged from the rest of the country, which supports Hypothesis 2. The figure distinguishes eleven states in the South in which the white population share was less than 80 percent from nineteen states in the North and Midwest with higher white population shares. Categorizing US states by racial composition enables comparison with the other world regions considered here, and it is identical to how some Americanist research categorizes the US South.Footnote ¹²⁸ The sample contains only states admitted to the Union prior to 1850, which yields a constant sample of states. In each year, a state can have up to four restrictions on voter eligibility, disaggregated by economic requirements, whites-only voting, poll taxes, and literacy requirements.Footnote ¹²⁹ Therefore, the dependent variable for the figure ranges between 0 and 4, and the lines represent averages between the two groups of states.

The earliest prevalent form of voter requirements (besides those based on gender) were property-holding and related economic requirements, although by 1850 few states retained them. Between 1850 and the beginning of Reconstruction in the late 1860s, the predominant form of voter restriction was race, which the Fifteenth Amendment (adopted in 1870) disallowed. But starting in 1890, when the federal government signaled it would not interfere with states’ voting practices, poll taxes and literacy requirements gained prevalence among southern states until the adoption of the Twenty-Fourth Amendment in 1964 and the Voting Rights Act of 1965 (the gray shaded area in Figure 6). These actions created a large gap in the average number of voter restrictions between southern states and the rest of the country—supporting Hypothesis 2—although several northern states featured literacy requirements until 1970 (Connecticut, Delaware, Maine, Massachusetts, New Hampshire, New York).

Figure 6

Voter Restrictions in US States