2. Colonial approaches to representative institutions

Although the simple, popular view of colonies is that they are areas ruled directly by some distant European power – often against the wishes of the inhabitants – the actual historical picture is much more varied and complex. Before 1960, colonies around the world had very different political institutions ranging from fully democratic governments with extended home rule to colonies with no representation that were directly ruled from Europe. The approaches both varied over time and between different colonial powers.

As Acemoglu et al. (Reference Acemoglu, Johnson and Robinson2001) have emphasised, Europeans who moved to the colonies may have had a preference for implementing representative political institutions. The parliament of Bermuda for example first met in 1620 and Barbados established its assembly in 1639, thereby preceding the establishment of most European parliaments. Both islands implemented property restrictions on voting known from England and would not abandon them before 1963 and 1950, respectively. Similar representative institutions had already been introduced in Virginia and would spread rapidly to all of the 13 British colonies in New England (Acemoglu and Robinson, Reference Acemoglu and Robinson2019; Congleton, Reference Congleton2011). Jamaica also introduced an elected assembly in 1677 while the Bahamas introduced elections and a small parliament on Royal order in 1729 (Encyclopedia Britannica, 2020).

This system spread in various guises to other British colonies. The tiny colony of the Cayman Islands held its first elections to a legislative assembly on the lines of Jamaican institutions in 1831 (Elections Office, 2020). In Africa, the Cape Colony held its first elections in 1854, in which voting was limited by property restrictions implemented regardless of ethnicity (Newton et al., Reference Newton, Benians and Walker1963). However, a more typical development in British colonies founded in the 19th century is exemplified by Rhodesia. In 1899, the colonial authorities introduced a Southern Rhodesia Legislative Council consisting of five members nominated by the British South Africa Company and four members elected by male British citizens subject to either a property or income restriction. Yet, the authorities ceded to popular pressure and the number of elected members of the council was gradually extended. As Rhodesia was granted self-government in 1923, the council turned into a Legislative Assembly that was elected by a franchise including women who were subject to the same property restrictions as male voters (Newton et al., Reference Newton, Benians and Walker1963).

A similar development over time was typical for most British colonies. The Legislative Council of the Gambia for example included a member elected in the colony area from 1944, besides the Governor, three ex officio members, three officials and six appointed members. By 1954, the Council included four elected members and the colonial authorities replaced it in 1960 with a House of Representatives that 2 years later consisted of five officials, three nominated members and 32 members elected in fully democratic elections (Bjørnskov and Rode, Reference Bjørnskov and Rode2020). Although the timing of these changes varied considerably from colony to colony and mainly appeared as a set of processes that was not coordinated by the colonial authorities in Westminster, they meant that almost all British colonies had representative political institutions in the late colonial period following WWII, albeit these differed in detail quite substantially.

The development in French colonies was markedly different and shaped by the particular approach to colonies in Paris. Although each British and Dutch colony was a separate administrative and political entity, the French approach was that all colonies were parts of France itself (Whittlesey, Reference Whittlesey1937), essentially making the relation of the political centre towards the territories into the defining element of domestic political development within all French colonies themselves.

Elections were introduced in all French colonies with the implementation of a new constitution for the Fourth Republic, which required all parts of France to elect members to parliament. For instance, this new constitution implied that Côte d'Ivoire in October 1945 elected two members of parliament, using two separate electoral colleges. One college was for citizens – essentially white French people – and another for non-citizens, meaning mostly Black Ivorians. The second college elected Félix Houphouët-Boigny who would go on to serve as a cabinet minister in the French government and later was elected the first president of an independent Côte d'Ivoire. Houphouët-Boigny achieved this election by exiling his main rival and would continue as president for the next 33 years (Encyclopedia Britannica, 2020). As such, he was a supreme example of the way the French – perhaps unwittingly – created a political elite with strong ties to France and its political establishment that Le Vine (Reference Le Vine1968, 372) calls ‘company boys’.

Elsewhere in the French colonies, the authorities introduced so-called Conseils Generale elected by ethnically stratified electoral colleges. In Dahomey and French Middle Congo – present-day Benin and the Republic of Congo – voters in a first French and predominantly White college elected 12 members to the Conseil while voters in a second, indigenous and Black college elected 18 members. The same type of institution was implemented in other French colonies, although with varying Conseil size. In Cameroon, voters elected 16 and 24 members in the two colleges while the Ivorian Conseil consisted of 18 and 32 members elected by the first and second college, respectively.

Certain French colonies with large shares of French citizens – often called Pied Noirs – were nevertheless given different institutions. By 1950, the Assemblee Algerien consisted of 60 Europeans and 60 Algerians elected on separate rolls while Tunisia's Grand Council consisted of 106 delegates, half of which were on a European roll and the other half on a Tunisian roll although Europeans only constituted about 10% of the population. Some of these delegates were appointed and the Muslim Tunisian delegates continued to be indirectly elected. Only the Jewish delegate was elected directly through a secret ballot while the French ballot was not secret and Tunisian voters were required to hold a high school degree.

However, the French institutional situation changed with the adoption in June 1956 of la Loi Cadre that introduced self-government in all colonies by direct mandate from Paris via territorial governments elected by a full franchise. The underlying idea was the creation of a relatively homogenous path to domestic democratic governance for all French colonies. These governments effectively paved the way for independence, which also provided the French authorities with strong incentives to manage their elections. Apart from the Ivorian example above, in which the opposition candidate was simply removed, a number of other examples of electoral ‘management’ were widely reported in the media at the time. The 1959 elections in Dahomey provided a pertinent example of the degree of electoral manipulation in many French colonies. Although the opposition Union Démocratique Dahoméenne obtained 44% of the vote, it only gained 11 of the 70 seats in parliament owing to the geographical distribution of the votes while the incumbent Parti Républicain Dahoméen gained 37 seats with 39% of the vote (Houngnikpo and Decalo, Reference Houngnikpo and Decalo2013: 286). Examples of outright fraud include the 1957 elections to the Ubangi-Shari Territorial Assembly in what is now Chad and the Central African Republic in which literally all seats went to the Mouvement pour l’Évolution Sociale de l'Afrique Noire. In the 1959 elections in what would become the Central African Republic, the party again won all seats and the country would remain a single-party regime until Jean-Bedel Bokassa's coup in December 1965. Similarly, Maurice Yaméogo had been part of the colonial administration for more than a decade before he joined a coalition government in 1957. After the death of the president of the council in 1958, Yaméogo took power and managed to impose a single-party dictatorship in Upper Volta (i.e. present day Burkina Faso) that was not opposed by the colonial authorities.

Other colonial powers pursued different policies towards their colonies in the late colonial period. Although Belgian colonies continued to be ruled directly and Portuguese colonies typically had appointed legislative councils, the Dutch approach was similar to British policy. All Dutch colonies moved to fully democratic political institutions after WWII. Although suffrage for example had been restricted to men paying taxes in Suriname, women could stand for office and a full franchise was introduced with the 1948 elections. Elections to the Estates of Suriname were as democratic as in the Netherlands and led to frequent changes of power even before Suriname was granted self-government in 1954 and final independence in 1975. The exception to the Dutch policy was Indonesia, in which 38 of the 60 members of the Volksraad were elected and the rest were appointed. Membership was further stratified across ethnic lines in a system similar to present day Lebanese political institutions.

In summary, colonial institutions and policies may not only have affected post-colonial constitutional norms but more general political norms defining how practical politics evolve. Colonial approaches differed substantially with French authorities taking a more active role in the creation of political institutions than their British or Dutch equivalents. The long period of elections to the Grand Council of French West Africa resulted in the establishment of party structures and ideologies in line with those in France. These parties represented a small French colonial elite, despite the fact that elections were fully democratic from 1947 onwards (Gonidec, Reference Gonidec, Becker, Mbaye and Thioub1997). The Grand Council for example included important independence figures such as the later dictator of Guinea, Ahmed Sékou Touré, who had also been a deputy to the French parliament, the later Ivorian dictator Felix Houphouët-Boigny, and Senegal's first minister of justice, Gabriel d'Arboussier, who happened to be the son of a French baron and colonial governor. The establishment of clear elite parties and institutions was thus aligned with the overall French approach of employing only French or French-educated civil servants (Whittlesey, Reference Whittlesey1937).

Conversely, it appears that British and Dutch development was far less controlled and planned. In the Caribbean, formalised political parties only arose in the 20th century, and a rich colony such as Bermuda had no formal parties before the 1963 elections. Elsewhere, the first leaders of several British colonies were far from representative of the colonial elite and often strongly unpopular with the colonial authorities. Both Kenya's first leader and long-time dictator Jomo Kenyatta and Botswana's much more democratically minded Seretse Khama had clashed with their respective colonial administrations but came to power in fully democratic elections (Bjørnskov and Rode, Reference Bjørnskov and Rode2020). These examples thus illustrate the potential long-term influence of colonial political institutions on subsequent political norms and accepted behaviour, if independent institutions continue colonial traditions of elite capture, management of elections or full respect for democratic outcomes.

3. Why would colonial institutions have such persistent consequences? A tentative interpretation of the relevant literature

Moving into the late colonial period after WWII, there were thus large differences between the different colonial powers, as well as across different colonies within the same empire. Not only did most colonies have representative political institutions – and some of the British colonies had had such quasi-democratic institutions for two centuries or more – but some of them were bona fide democratic. By 1959, immediately before the decolonisation wave of Africa, 64 of the 85 colonies covered by the Bjørnskov and Rode (Reference Bjørnskov and Rode2020) database in that year had representative political institutions. Only the Central African Republic had turned to a single-party regime at that time while 26 of the colonies were fully democratic.

The question whether the differences in political institutions in the late colonial period continue to shape institutions in the 21st century is essentially one of institutional persistence and path dependence – how important are historical institutions to the modern status? – which is the question asked by a notable strand of development research in recent years. Most empirical contributions investigate the impact of colonial structures on comparative economic development (e.g. Acemoglu et al., Reference Acemoglu, Johnson and Robinson2001; La Porta et al., Reference La Porta, Lopez-de-Silanes and Shleifer2008), while a subset has more specifically analysed the importance of colonialism for the democratic development of former colonial dominions (e.g. Guardado, Reference Guardado2018; Lee and Paine, Reference Lee and Paine2019; Olsson, Reference Olsson2009). Both find overwhelming evidence that colonial history matters a great deal for the development of formal institutions and economic prosperity after transition to independence, but this literature is often rather ambiguous with respect to how exactly path dependence and institutional persistence matters (Boettke et al., Reference Boettke, Coyne and Leeson2008; Maseland, Reference Maseland2021).

A growing body of related literature has since focused extensively on the factors that determine successful (or unsuccessful) processes of institutional transfers during the colonial era, attempting to uncover how exactly history matters for post-colonial institutional outcomes. This body of literature highlights a common set of factors that determine successful institutional transplants, such as the importance of individual leaders (Couyoumdjian, Reference Couyoumdjian2012; Couyoumdjian and Larroulet, Reference Couyoumdjian and Larroulet2018), historical institutional precedents (Berkowitz et al., Reference Berkowitz, Pistor and Richard2003; Boettke et al., Reference Boettke, Coyne and Leeson2008; Pavlik and Young, Reference Pavlik and Young2020, Reference Pavlik and Young2021), the compatibility of formal and informal institutions (Berkowitz et al., Reference Berkowitz, Pistor and Richard2003; Boettke et al., Reference Boettke, Coyne and Leeson2008; Gutmann and Voigt, Reference Gutmann and Voigt2020; Seidler, Reference Seidler2014, Reference Seidler2018), as well as the factor time, where recipients can experiment around and adjust the imported institutions to local practices (Seidler, Reference Seidler2018). Obviously, these factors are not mutually exclusive and successful institutional transplants to former colonies often present a combination of several elements. Finally, questions of the persistence of colonialism for current institutions have only been raised very recently, with some authors emphasising its declining impact and the growing importance of pre-colonial institutions (Maseland, Reference Maseland2018).

Despite these important advances, as a whole, this literature remains quite ambiguous on the exact transmission mechanisms that make successful institutional transfers ‘stick’ (cf. Boettke et al., Reference Boettke, Coyne and Leeson2008), at least in a strictly empirical sense. In some way, the present state of the applied literature offers two possible interpretations: first, a number of notable contributions examine whether the differences in political institutions that can be observed in the late colonial period have consequences for subsequent political development in independent states, and if they characterise its political institutions of the 21st century? In one of the most influential contributions to this literature, Acemoglu et al. (Reference Acemoglu, Johnson and Robinson2001) argue that the introduction of representative political institutions in European colonies was much more likely when Europeans chose to settle permanently: where the disease environment was favourable, European immigration brought the beginnings of representative democratic institutions. Their claim has since been re-evaluated in the relevant literature (e.g. Ketterer and Rodríguez-Pose, Reference Ketterer and Rodríguez-Pose2018), albeit with some criticisms voiced regarding the accuracy of their data and strategy, and the extent to which the claim is specific to a certain class of colonies (e.g. Albouy, Reference Albouy2012; Fails and Kriekhaus, Reference Fails and Krieckhaus2010). Olsson (Reference Olsson2004) for example notes that the two different waves of colonisation had different ideological backgrounds and therefore different consequences for the institutions transplanted in the colonies.

Since Lipset's (Reference Lipset1959) seminal study of democratisation, a different tradition has also focused on the role of education. Lipset believed that some level of literacy and education is necessary for democracy to be a viable political institution, as voters would otherwise either not be informed about political discussions or unable to understand their consequences. Again, colonies in the late colonial period varied markedly in citizens' access to education. Although a number of colonies only offered primary education, the colonial powers in general invested in education in the 20th century although the quality of the educational infrastructure varied substantially. Parker (Reference Parker1959) for example notes the well-known quality of educational institutions in Rhodesia in the post-war period, which were attended by 85% of Black African children. Conversely, only 3.3% of children finished school in French West Africa around 1950 (Diop, Reference Diop, Becker, Mbaye and Thioub1997). Nevertheless, some colonies both offered high school degrees and higher education.Footnote ¹ The University of Dakar for example grew out of a series of French institutions in 1957 and the University of the West Indies was founded in Jamaica 9 years earlier. Elsewhere, reputable universities in Europe established franchises such as Codrington College in Barbados, which for decades prior to independence provided university education courtesy of Durham University in the United Kingdom (Statesman's Yearbook, 1950–1954). Following the tradition from Lipset (Reference Lipset1959), colonial educational decisions might therefore affect democratic development after independence.

Despite these very different approaches to the roots of colonial and post-colonial democracy, all of these studies have a common underlying thread: they potentially see colonial political institutions as mediators of the relation between colonial characteristics and post-colonial outcomes. The basic idea is that the institutional characteristics of the colony largely determine post-colonial outcomes. Factors like demography, education and culture are only important in the sense that they determine the characteristics of the individual colony, which will in turn determine the characteristics of the post-colonial independent state. So in this first scenario, transplanted institutions are effective, but certain characteristics determine whether good-quality transplanted institutions are likely. Interestingly, due to a lack of adequate data, the mediation argument is never explicitly tested, but rather implicitly assumed.

Second, another strand of this literature essentially assumes that certain features, such as the size of the European population, institutions of higher education and total population, affect the probability of having representative political institutions prior to independence in the late colonial period, and that the influence of those same factors persist in the sense of affecting the choice of political institutions today. For example, Hariri (Reference Hariri2012: 471) argues that countries with established statehood prior to colonisation were less likely to experience institutional transplants, as they had ‘enough state infrastructure that the colonial powers would rule to a considerable degree through existing institutions’. Some of these institutional transplants may have enabled the development of democracy while others have had the opposite effect. In addition, it remains possible that some of the transplants of constitutional norms and institutions were not necessarily formal and immediately measurable. Instead, a likely way that historical factors as well as different colonial approaches may have affected subsequent development can be through the establishment of constitutional norms or culture – i.e. the unwritten rules that effectively constrain political actors (cf. Ferejohn, Reference Ferejohn1986; Seidler, Reference Seidler2014; Voigt, Reference Voigt2020).

In this second strain of the literature, colonial and postcolonial political institutions are jointly determined by the same underlying set of variables (education, demography and culture). The institutional characteristics of colonies obviously matter for post-colonial outcomes, but factors like demography, education and culture also determine the characteristics of the post-colonial independent state to a substantial degree, and in a parallel manner. In this second scenario, transplanted institutions do not play an overly important role, rather it would be the underlying characteristics of the country that determine its current level of democracy.

Summarising, the empirical literature to date does not distinguish between these two scenarios of how institutional transplants might affect post-colonial outcomes. Although making the associated arguments, these studies typically do not measure late colonial political institutions, because they often lack data about the potentially mediating factors. To some degree, the different scenarios highlighted above also reflect an older debate between Acemoglu et al. (Reference Acemoglu, Johnson and Robinson2001, Reference Acemoglu, Gallego and Robinson2014) and Glaeser et al. (Reference Glaeser, La Porta, Lopez-de-Silanes and Shleifer2004) on the structural determinants of long-run growth, where Maseland (Reference Maseland2021) recently highlights that it is still largely unclear how ‘deep determinants’ influence present economic outcomes. What we introduce in the following, is a measure of late colonial political institutions and their potential determinants, that further allows us to test whether the influence of colonial social characteristics is direct or indirect, i.e. jointly determining or mediating.

4. Data and sources

The empirical application relies on several data sources: first, we employ the recently developed database in Bjørnskov and Rode (Reference Bjørnskov and Rode2020) who provide an update and expansion of the information on regime types and political institutions in the Democracy-Dictatorship data by Cheibub et al. (Reference Cheibub, Gandhi and Vreeland2010) dating back to 1950. Among other innovations, Bjørnskov and Rode (Reference Bjørnskov and Rode2020) provide new institutional data for periods under colonial rule covering more than 90 entities at present, thereby offering substantially wider coverage of non-sovereign countries than comparable alternatives, such as the Varieties of Democracy project (Coppedge et al., Reference Coppedge, Lindberg, Skaaning and Teorell2016).

As part of the post-1990 wave has been the rise of illiberal democracy, where some states have introduced multi-party elections but implemented in a way that de facto implies little electoral risk for the incumbent (Zakaria, Reference Zakaria1997), it might sometimes be difficult to distinguish between these two concepts. For that reason, Bjørnskov and Rode (Reference Bjørnskov and Rode2020) provide a variable that captures whether colonies have no regular elections, hold elections in one-party states, elections with opposition parties but without an actual chance of government change or have full electoral democracy. For the present analysis, this allows us to separate colonies into three groups: those with a functioning democracy, territories in which the colonial power still ‘directed’ the elections and colonies without any representative institutions. Of 94 former colonies in the dataset, 58 are currently categorised as democracies, 34 as electoral autocracies and only two do not have multi-party institutions: Equatorial Guinea is essentially a single-party military dictatorship and Somalia, as the quintessential failed state, is unable to even hold nation-wide elections. From the same source, we add a dummy capturing if a colony was self-governing.

Second, we combine these data with hand-collected information from early 1950s editions of The Statesman's Yearbook. This reference book has been published annually since 1864 and provides general information on the countries of the world, among it, reliable data on educational institutions and population structure at the time. To the best of our knowledge, the only other recent use of this source in political science or economics is by Bjørnskov and Rode (Reference Bjørnskov and Rode2020). In particular, we hand-collect data on the relative size of the European (or White) population at the end of the 1940s, the total population, the total land area covered by the colony and whether colonies presented institutions of secondary or tertiary education at the time (Statesman's Yearbook, 1950–1954).

Third, in further robustness tests we employ additional control variables from other sources, such as the famous settler mortality rates by Acemoglu et al. (Reference Acemoglu, Johnson and Robinson2001), which we introduce to control for historical population patterns. Similarly, we employ a variable from Pavlik and Young (Reference Pavlik and Young2020, Reference Pavlik and Young2021) that captures medieval and early modern assembly experiences to capture variation in historical political institutions. We also code a dummy variable capturing whether a colony had representative institutions prior to 1900.Footnote ² All variables are summarised in the Appendix, which also includes a list of all colonies included, with their modern names.

With a cross-sectional dataset of up to 65 former colonies that are all observed before and after their transition to independence, we attempt to empirically establish the determinants of countries' political institutions at the moment of independence, and for the present day, in the following section. Apart from applications by Bjørnskov and Rode (Reference Bjørnskov and Rode2020) and Lee and Paine (Reference Lee and Paine2019), we are also unaware of any further study that directly associates pre-independence political institutions of colonies with the democratic outcomes of the post-independence period. Of those, the current study is the only one that tests whether the influence of colonial social characteristics is jointly determining, or mediating.