THEORY

A central line of inquiry in the quantitative political science and economics literature on colonialism examines why colonizers may adopt direct or indirect rule, and what these colonial subtypes mean for the persistence of precolonial domestic power structures. Gerring et al. (Reference Gerring, Ziblatt, Van Gorp and Arévalo2011) attribute indirect colonial rule to the existence of a precolonial state. According to their thesis, precolonial polities with greater levels of “statedness,” that is, formal state institutions, were more likely to be indirectly ruled after occupation. Imposing direct rule on highly institutionalized states is unlikely due to the high cost of replacing the precolonial state altogether. Müller-Crepon (Reference Müller-Crepon2020) also finds that more centralized states in Sub-Saharan Africa were more likely to be indirectly ruled, and that indirect rule more frequently preserved precolonial ruling dynasties than direct rule. Recent work on the legacies of precolonial states and conflict also links elite groups that organized as precolonial states with the incidence of postcolonial conflict, albeit with divergent findings (Dincecco et al. Reference Dincecco, Fenske, Menon and Mukherjee2022; Paine Reference Paine2019; Wig Reference Wig2016). Given that precolonial and colonial institutional legacies have been shown to predict a range of political and economic outcomes, understanding precisely which elites populate those institutions over time can shed light on why.

Yet, interdisciplinary scholarship on indirect rule is replete with evidence of colonial reengineering of the precolonial political order (Lee Reference Lee2017; Naseemullah and Staniland Reference Naseemullah and Staniland2016; Slater Reference Slater2010). Traditionalization is one such method, where the colonial administration revised precolonial institutions and hierarchies under the guise of legitimating their rule and increasing control (Boone Reference Boone1995; Reference Boone2014; Crowder Reference Crowder1964; Herbst Reference Herbst2000; Hobsbawm and Ranger Reference Hobsbawm and Ranger2012; Mamdani Reference Mamdani1996). Scholars argue that indirect rule used the legitimizing screen of precolonial institutions to shift the precolonial distribution of power to favor colonial interest. Much of this work focuses on sub-national local elites and institutions as a means to control peripheral regions (Baldwin Reference Baldwin2015; Banerjee and Iyer Reference Banerjee and Iyer2005; Mukherjee Reference Mukherjee2021).

Our theoretical intervention helps to reconcile these accounts of persistence and change by addressing why and how colonizers shift the composition of the national-level domestic political class within nominally continuous state institutions. To do so, we turn to insights from scholarship on authoritarian power-sharing. This literature offers two key insights to studying the redistribution of power in colonial contexts. First, scholars of authoritarianism argue that power-sharing with elites emerges to mitigate their threat of revolt and ensure regime survival, and that power-sharing may be institutionalized in order for the dictator to credibly commit to the elites and undercut challengers (Boix and Svolik Reference Boix and Svolik2013; Bueno de Mesquita et al. Reference Bueno de Mesquita, Smith, Siverson and Morrow2003; Gandhi and Przeworski Reference Gandhi and Przeworski2007; Meng Reference Meng2019; Svolik Reference Svolik2012). We argue that colonial regimes, like sovereign autocracies, face significant threats of revolt from the populations that they govern, especially elites. To counteract the threat of anti-colonial revolt from the domestic precolonial elite, indirect rule’s use of precolonial institutions to share power with domestic elites is analogous to institutionalized power-sharing in sovereign autocracies.

The second transferable insight from the authoritarian power-sharing literature to the study of indirect rule is Boix and Svolik (Reference Boix and Svolik2013)’s concept of “total power.” We argue that this is a useful heuristic for conceptualizing the institutionalization of power-sharing between the colonizer and domestic elites. In a given precolonial polity, total power is distributed among domestic elites. Colonialism disrupts the precolonial distribution of total power, as the colonizer captures a significant share of total power, which is (exogenously) determined by the metropole’s willingness to invest in the colony.Footnote ¹ The colonizer thus forms a ruling coalition with the minimum number of domestic elite allies—the “collaborators”—in order to achieve the threshold share of total power that is needed to rule. Under indirect rule, the colonizer institutionalizes power-sharing with domestic collaborators, in order to mitigate the threat of rebellion, via the creation of new colonial institutions, or the preservation of precolonial institutions, including coercive, executive, and legislative institutions.

The logic of authoritarian power-sharing helps explain why the colonizer may institutionalize power-sharing with domestic elites under indirect rule. However, this logic does not directly address our central question of why and how the colonizer would change the precolonial power distribution within precolonial elites, conditional on the share of power controlled by the colonizer. This literature generally conceives of power-sharing between an autocrat and a homogeneous elite. In contrast, our framework assumes that there are two classes of the precolonial elites who are heterogeneous with respect to the threat of revolt they may pose to the colonizer, and that each class holds sufficient share of power that would make it a viable ally for the colonizer.Footnote ² We expect the colonizer will select the most politically congruent elite class—with the least threat of anti-colonial rebellion—as collaborators in the national-level colonial ruling coalition (political congruence). Put differently, the colonizer will redistribute power away from the most politically oppositional elites, and toward the elites that are most politically congruent with the colonizer’s objective of regime survival. This happens in order to mitigate frictions within the coalition on the question of regime survival. The excluded oppositional elite class will not be able to stage a successful anti-colonial rebellion without the support of some collaborators, as long as the colonizer is sufficiently strong relative to the collaborators, and the colonial coalition is sufficiently strong relative to the excluded oppositional elite class (Boix and Svolik Reference Boix and Svolik2013).

Regime survival—or mitigating the threat of revolt—does not fully capture colonial incentives to redistribute power. This reflects a crucial difference between sovereign autocracies and colonial regimes. In sovereign autocracies, the dictator’s tools of repressing an elite rebellion are limited to the domestic state resources, and in the case of a successful rebellion the dictator has little chance of exit. In colonial regimes, the colonizer has a higher chance of repressing elite rebellion by drawing on extraterritorial power (e.g., financial, military) from the metropole. If a rebellion succeeds, the colonizer may exit with relative ease. Given the weaker rebellion threat in colonial settings, we further extend the authoritarian power-sharing model by arguing that colonial power-sharing can be explained by the threat of the domestic elite to undermine the colonizer’s broader set of objectives beyond regime survival. Institutionalized power-sharing under indirect rule would thus mitigate the threat of undermining the strategic colonial objectives. We assume that domestic elites are heterogeneous with respect to their threat of undermining colonial objectives.

A wide array of objectives may drive colonial expansion. Economic extraction is often cited as a motivation for late colonialism (Acemoglu, Johnson, and Robinson Reference Acemoglu, Johnson and Robinson2001; Dell and Olken Reference Dell and . Olken2020; Robinson and Acemoglu Reference Robinson and Acemoglu2012). Some have focused on the cultural and ideological drivers of imperialism, such as a supremacist “civilizing” mission or religious proselytism (Daughton Reference Daughton2006; Porter Reference Porter1992). Other strategic goals of colonial rule include maintaining access to trade routes, securing borders, and inter-state competition (Lange, Mahoney, and Hau Reference Lange, Mahoney and Hau2006; Mahoney Reference Mahoney2010; Paine, Qui, and Ricard-Huguet Reference Paine, Qui and Ricard-Huguet2024). While economic extraction prevails as the most common objective in the late colonial period (Beckert Reference Beckert2014; Ferro Reference Ferro1997), our theory applies to other noneconomic objectives that may motivate a colonizer to occupy a polity. Across the range of motivations, we expect the colonial autocrat will opt to redistribute power toward the elite that is most congruent with (least oppositional to) a broader set of colonial objectives (strategic congruence).

Our theory has two main implications. First, we expect to see an increased share of total power for congruent elites relative to oppositional elites under indirect rule at the national level. Whether the congruent elite would hold a greater absolute share in total power relative to their share in the precolonial period is ambiguous due to the fact that the colonizer controls a significant share of total power. Second, we argue that the redistribution of national-level power will reflect the geography of realizing the colonizer’s objective. We expect that the power redistribution toward congruent elites (and away from oppositional elites) will be concentrated in regions related to the colonizer’s primary objective. The effects of economically-motivated occupation, for example, should be most prevalent in regions that generate the most economic surplus. In the analysis that follows, we refer to this spatial dimension of colonial interest as colonial exposure. While this effect may be a consequence of intentional, targeted intervention by the colonial administration, it is also possible that this redistribution emerges as an unintended consequence of policies designed to achieve the colonial objectives.

The findings of Gerring et al. (Reference Gerring, Ziblatt, Van Gorp and Arévalo2011) imply a contending, alternative explanation that by-passes our theory of elite congruence. Given that precolonial state institutions are more likely to persist under indirect rule, it is also possible that we would simply see either the maintenance or expansion of power in favor of precolonial elites that already control the state. The path dependant intuition of this argument is straightforward, as it would involve less colonial intervention to simply preserve the power of incumbent state elites. On the aggregate level, this alternative explanation would predict that precolonial state elites persist and maintain their relative power under the colonial regime, even though their absolute power would likely decrease due to colonization. Sub-nationally, precolonial state capacity, rather than elite congruence, would be the driving factor behind any colonial redistribution of power in favor of incumbent power elites. If this alternative explanation is indeed the one that captures the most variation in colonial redistribution of power, measures of precolonial statedness should be more predictive than measures of colonial exposure.

In the next section, we provide historical context related to our theory in the case of Egypt. We describe the power-sharing arrangement between the precolonial Khedival regime and the two most powerful elite groups: the LE and the RMC. We set the stage for our empirical exercise by demonstrating the colonial regime’s economic motivations to increase Egyptian cotton production and reflect on what previous work has shown about the distribution of domestic political power among precolonial elites under British indirect rule.

HISTORICAL BACKGROUND

The remainder of this article focuses on the case of precolonial and colonial Egypt. Egypt had a well-documented parliament before and after the British occupation, allowing us to observe changes in the distribution of total power in the national-level domestic political class below the executive more extensively than in cases without a parliament. In this section, we describe the evolution of power-sharing in precolonial and colonial Egypt with the LE and the RMC.

Egypt was an autonomous Ottoman vassal state from 1805 to 1882, a de facto British colony under nominal Ottoman sovereignty from 1882 to 1914, and a British protectorate from 1914 to 1922, when it gained nominal independence in February 1922. During the precolonial and colonial periods, Egypt’s domestic government was ruled by the dynasty of the Ottoman viceroy Muhammad Ali (1805–48), whose descendants adopted the title of “Khedive” between 1867 and 1914. By the second half of the 19th century, the Khedival regime institutionalized power-sharing with two elite groups: the LE and the RMC (Abbas and El-Dessouky Reference Abbas and El-Dessouky2011; Sayyid-Marsot Reference Sayyid-Marsot1969; Schölch Reference Schölch1974).

The Khedival regime institutionalized power-sharing with the LE by awarding them leadership positions in government. The majority of Egypt’s precolonial ministers, provincial governors, top military officers, and chiefs of government agencies came from the LE before the British occupation in 1882 (Collins Reference Collins1984). The Khedival family were the largest landowners in Egypt and used state-owned usufruct land to grant large landholdings to these officials (Baer Reference Baer1962; El-Dessouky Reference El-Dessouky1975). The earliest 19th century LEs were ethnic Turco–Circassians (Abbas and El-Dessouky Reference Abbas and El-Dessouky2011; Baer Reference Baer1962), but historians agree that intermarriage between Ottoman and Egyptian upper-class families, as well as upward class mobility for certain Egyptians, resulted in an ethnically blended elite by the 1870s (Sayyid-Marsot Reference Sayyid-Marsot1984) who were defined by their out-sized wealth and access to executive decision-making (Blaydes and El Tarouty Reference Blaydes, Tarouty, Cammett and Jones2022).

Political histories of Egypt are clear that the RMC enjoyed institutionalized power-sharing under the precolonial Khedives, first at the local level and later at the national level. The RMC were predominantly comprised of village headmen. The village headman role predates Khedival rule, but became incorporated into the Khedival local bureaucracy after the abolition of tax farming in 1813 (Cuno Reference Cuno1992). As intermediaries between the state and the peasantry, headmen fulfilled a hybrid role as mayor-tax collectors under the Ministry of Interior. Brown (Reference Brown1990, 29) defines the RMC as a class of commercial farmers “whose presence extended throughout rural Egypt.” The RMC were distinguished from the peasantry by the fact that they could afford paid (or slave) labor to capitalize on their agricultural production and owned landholdings that fell in the middle of the distribution (Brown Reference Brown1990). Davis (Reference Davis1983, 40–1) notes that the RMC also developed a distinct class consciousness through a shared material interest (i.e., capitalist cultivation of cash crops and being subjected to heavy taxation), and recruitment into positions in the lower ranks of government bureaucracy and military.

The precolonial parliament (1824–82) institutionalized national-level power-sharing between the Khedive and the RMC. Sayyid-Marsot (Reference Sayyid-Marsot1984) argues that the earliest parliament (al-Majlis al-‘ali, 1824–37) was created with the primary goal of supporting Muhammad Ali’s rural reform programs. As a result, the RMC held the majority of seats by design. Members of the 1866–82 parliament, Majlis shura al-nuwwab in 1866–79 and Majlis al-nuwwab al-misry in 1881–82, were mostly elected village headman, while the LE predominantly held appointed positions, such as Speaker of the Chamber (see Figure 3).

By the late 1870s, precolonial RMC MPs began to demand greater power-sharing and legislative oversight. Historians link the growth in RMC power to an abrupt increase in their economic power due to a boom in Egypt’s cotton exports (Cuno Reference Cuno1992). Egypt was a prominent exporter of high-quality long-staple cotton since 1820, but a cotton boom during the United States’ civil war led Egyptian cotton exports to quadruple (Owen Reference Owen1969). The blockade of Southern cotton trade meant that industrializing Britain needed a new source of raw materials for its textile mills (Cole Reference Cole1993, 58). The cotton boom enriched both the LE and the RMC, although only the LE enjoyed the legal right to force local peasants to work on their large estates (Abbas and El-Dessouky Reference Abbas and El-Dessouky2011). These restrictions on local labor induced the RMC to purchase slaves from Sudan (Cuno Reference Cuno2009; Helal Reference Helal1999; Saleh Reference Saleh2024).Footnote ³

On the eve of colonial rule, the RMC acquired a greater share of total power than ever before. Key historical events are summarized in Figure 1. The Egyptian government’s debt default in 1876 increased RMC dissent. The Khedival regime borrowed heavily from European powers to finance domestic infrastructural development. Britain and France, as the primary stakeholders, established a system of dual control over Egyptian finances through a new institution, the Caisse de la Dette Publique, and gained ministerial appointments in finance and public works. The parliamentary debates from this period feature the RMC’s opposition to increased European intervention in Egyptian domestic affairs and worsening economic conditions (Dar al-Watha’iq al-Qawmiya 2017). The British used their increased influence to force Khedive Isma’il to abolish slavery and announce the future emancipation of slaves via the 1877 Anglo–Egyptian Slave Trade Convention. In response, the RMC MPs began to demand legislative powers and oversight over the budget and ministerial cabinet (Dar al-Watha’iq al-Qawmiya 2017). These demands were so robust that they materialized in a draft constitution in 1879 (Subhi Reference Subhi1947). To quell dissent, the Khedive shuttered parliament and European powers colluded to replace Khedive Isma’il with his son, Tewfiq. By 1879, the RMC was strong enough to credibly push for greater power-sharing and enhanced their strength by allying with the ‘Urabi movement.

Figure 1.

Political and Legislative Event Timeline (1879–82)

The ‘Urabi movementFootnote ⁴ was led by Colonel Ahmed ‘Urabi, an Egyptian military officer from a RMC background (Cole Reference Cole1993, 207). The 1876 default and austere fiscal regime led to the drastic reduction of the size and budget of the military, creating a discontented pool of unemployed military officers (Cole Reference Cole1993). Tapping into growing resentment against European intervention and economic turmoil, ‘Urabi’s nationalist movement spread throughout Egypt and called for sovereign control over the Egyptian treasury, parliamentary oversight of the Khedive, and increased representation of non-LEs in the Egyptian executive and military officials. The threat of overthrow from RMC MPs and allies in the military-led ‘Urabi movement was strong enough to force Khedive Tewfiq to defy the British and French and to unilaterally reopen the parliament in 1881 (Al-Rafi‘i Reference Al-Rafi‘i1949). In 1882, ‘Urabi became the first RMC minister when the Khedive appointed him as the Minister of War. The final precolonial parliament (1881–82) passed a new constitution that expanded the franchise and established the legislature’s authority over the executive (Dar al-Watha’iq al-Qawmiya 2017; Subhi Reference Subhi1947). The new constitution was ratified by Khedive Tewfiq into law on February 7, 1882.

Historians argue that the European powers feared what the success of ‘Urabi and the RMC meant for their interests in Egypt. According to Sayyid-Marsot (Reference Sayyid-Marsot1969, 17), it was clear to everyone—Egyptian and foreign —that the ‘Urabists (including the RMC) were the only power in the country in early 1882; the British Consul in Alexandria reported that the Khedive was “powerless,” and that the Anglo-French influence was, “steadily decreasing. We can only regain our ascendancy by the destruction of the military supremacy.” Britain’s financial risk under the new status quo was threefold. First, a parliament with powers of budgetary oversight and controlled by RMC interests threatened Britain’s likelihood of recouping Egypt’s outstanding debt (Jakes Reference Jakes2020, 1). Second, British manufacturing demanded cheap cotton for textile manufacturing in the metropole (Owen Reference Owen1969; Schölch Reference Schölch1976). Third, the British believed that a change in the balance of power would jeopardize Britain’s trade routes should they lose their strategic, preferential access to the Suez Canal (Sayyid-Marsot Reference Sayyid-Marsot1969), but Galbraith and Sayyid-Marsot (Reference Galbraith and Sayyid-Marsot1978) argue that this concern was a distant third to Egyptian debt and access to cotton. On July 11, 1882, the British launched their occupation of Egypt.

Under the British, Egypt became “huge cotton plantation to satisfy the needs and desires of a colonial power” (No Author 1964). The area under cotton cultivation expanded from 693,000 feddansFootnote ⁵ in 1882 to 1,723,094 feddans in 1913 (Owen Reference Owen1969, 186). Early colonial expansion of cultivation and irrigation focused on the cotton-producing provinces of Lower Egypt and only expanded to Upper Egypt in the early 1900s (Abul-Magd Reference Abul-Magd2013; Tignor Reference Tignor1966). Cotton was so central to colonial Egypt that the state-owned railway lost 26,500 Egyptian pounds in 1888 due to the “smallness of the cotton crop,” which reduced shipments from the provinces to the port in Alexandria (Parliament Command Papers 1890, 9).

The question of how Britain’s economic interests in Egypt may have altered the elite composition of Egyptian parliament remains unanswered in the historiography of Egypt. There are several reasons for this. The first is that historical and political studies of the period often favor the use of certain colonial terminology that blurred class distinctions, therefore making it difficult to track continuity and change by social class. In some sources, the RMC are considered fellaheen (peasantry), while in others they are grouped with the LE as “notables.” This elision of terminology means that many historical studies assume continuity in the precolonial and colonial ruling elite of “notables,” without considering how British rule affected the relative shares of power between the LE and RMC at the national level. The second reason why this question has received less attention is due to a long-held assumption that the parliament itself was “powerless.” Not only has this assumption been challenged by advances in the political science literature on authoritarian parliaments, but also by scholarship on Egypt’s colonial experience that identifies the parliament as an important arena of anti-colonial political opposition, particularly in the years leading up to the 1919 Revolution (Tignor Reference Tignor1976).

DATA

To test our theory in the Egyptian case, we constructed a dataset at the MP-session level that spans the universe of Egyptian MPs from the first precolonial parliamentary session under Muhammad ‘Ali Pasha in 1824–37 to the promulgation of the first postindependence constitution in 1923. This includes 11 parliamentary sessions: five sessions during the precolonial period, and six under colonial rule.Footnote ⁶ Our data are based on a primary source in Arabic, History of Parliamentary Life in Egypt since the Era of Muhammad ‘Ali Pasha, that was compiled by Subhi (Reference Subhi1947) from MP lists in the Egyptian parliamentary archives.

For each individual MP, we know their full name, occupation (e.g., village headman), and honorific title (e.g., pasha, bey, effendi, sheikh). We also know whether each MP is elected or appointed, their date of entry into the parliament, their constituency at the province, district, or village level, and whether they held an executive position (e.g., speaker of parliament). We organize the data by session and include the dates of parliamentary sessions and the official name of each chamber.

We manually matched MPs across sessions and created a unique identifier for each MP using the MP’s full name.Footnote ⁷ We also created a unique dynasty identifier that traces family names across MPs and sessions, where we define the family name as the last name of an MP’s full name. We assigned MPs to provinces—the level at which parliamentary constituencies were defined during the colonial period (1883–1923) (see Section A1 of the Supplementary Material for details). The final dataset consists of 771 unique MPs, who served for a total of 1,102 MP-session observations, spanning the period from 1824 to 1923.

Our main outcome variable is the social class origin of each MP. We classified MPs into three classes based on three variables: honorific title, occupation, and the urban/rural status of their constituency (see Section A2 of the Supplementary Material for the historical basis for our coding criteria). We used the initial honorific and occupational titles that are observed in each MP’s first session in parliament.Footnote ⁸ We then assigned each MP to a social class following the steps described in Section A2 of the Supplementary Material and summarized here. The LE (289 MP-session observations) consist of pashas and beysFootnote ⁹ —the highest honorific titles in Khedival Egypt—and top bureaucrats. The RMC (679 MP-session observations) consist of MPs in rural constituencies, with non-missing honorific titles (excluding pashas and beys) or non-missing occupational titles (except top bureaucrats). The urban middle class (57 MP-session observations) consist of MPs in urban constituencies, with non-missing honorific titles (excluding pashas and beys) or non-missing occupational titles (except top bureaucrats). There are 77 MP-session observations with missing social class, because they are either not assigned to a constituency (and are not pashas, beys, or top bureaucrats), or are assigned to a constituency yet their honorific and occupational titles are both missing.

Our main explanatory variable is precolonial cotton productivity at the province level—the level at which we observe parliamentary constituencies of MPs. Precolonial cotton productivity measures colonial exposure, or the potential contribution of each province to Egypt’s economic surplus during the colonial period. Cotton productivity is measured using the average cotton yield per feddan in 1877 in each province as reported in Egypt’s 1877 Statistical Yearbook (Ministère de l’Intérieur 1877). We also control for precolonial cereal productivity (the average yield of wheat, barley, and beans per feddan by province in 1877) using the same source.Footnote ¹⁰ Both cotton and cereals productivity measures are continuous. Figure 2 shows the spatial distribution of cotton and cereals productivity across provinces. On average, Lower Egypt (Nile Delta) produced more cotton and cereals than Upper Egypt (Nile Valley), but there was significant variation in productivity within each region.

Figure 2.

Cotton and Cereals Yield Per Feddan in 1877

Note: The maps show the province-level distribution of cotton and cereals productivity in 1877. Cotton productivity is the cotton yield in qintars per feddan, and cereals productivity is the yield of wheat, barley, and beans in ardabbs per feddan, where 1 feddan = 6,368 square meters, 1 qintar = 44.5 kilograms, and 1 ardabb = 135 kilograms. We use the 1882 population census administrative divisions (CAPMAS 2008). The quartiles of precolonial cotton productivity in 1877 are defined based on the cross-province distribution: $ {Q}_1=0 $ , $ {Q}_2=1.067931 $ , $ {Q}_3=1.756632 $ . Source: Ministère de l’Intérieur (1877).

Figure 3.

The Social Class Composition of Members of Parliament, 1824–1923

Note: See the “Data” Section and Section A2 of the Supplementary Material for details about the classification of MPs into the three social classes.

Table A5.14 in the Supplementary Material shows the summary statistics in our MP-session dataset during the precolonial period (1866–82), broken down by the level of precolonial cotton productivity in the province in 1877 (above and below the median). During the precolonial period, we fail to detect statistically significant differences in the social class composition of MPs from provinces with higher precolonial cotton productivity and those from lower cotton productivity provinces. Examining the component variables of social class, we also fail to find statistically significant differences with respect to these variables in 1866–82.

In the next section, we empirically test the effect of the British occupation on the distribution of political power in the Egyptian parliament, and the degree to which this redistribution is consistent with our theory of elite congruence and colonial exposure.

EMPIRICAL ANALYSIS

Our theory predicts that the colonizer would redistribute political power in favor of domestic elites who are more congruent with colonial objectives. In Egypt, we argue that the colonial government’s goals were not only regime survival, but also to maximize Egypt’s economic surplus to ensure debt repayment and maintain access to Egyptian cotton as a primary input in British manufacturing. We expect that redistribution of power would be most profound for MPs representing regions with greater colonial exposure. In Egypt, high cotton-producing provinces were the most exposed to both colonial regime survival and economic objectives because they produced most of Egypt’s economic surplus.

We begin our analysis with historical evidence from archival and secondary sources that shows that the LE were the most congruent with the British in both the political and economic domains relative to the RMC. We then quantitatively show that the redistribution of power toward the LE and away from the RMC was greater in cotton-producing provinces, areas that were most exposed to colonial economic objectives. To explore the mechanisms driving this relationship, we demonstrate that within cotton-producing provinces, the colonial redistribution of power was greater in provinces where (1) the RMC were more politically oppositional, and (2) the LE were more economically congruent, and the RMC was more economically oppositional, before colonialism. Finally, we demonstrate how the colonizer reengineered the parliament to favor the LE.

Colonial Exposure and Redistribution of Parliamentary Power

The LE were more politically and economically congruent with Britain than the RMC. The British administration was well-aware that the RMC was the “backbone of the ‘ArabistFootnote ¹¹ party,” and had demonstrated their capability to mobilize against the British administration (Baring Reference Baring1908, 187), both in parliament and across Egypt (Cole Reference Cole1993).Footnote ¹² RMC MPs had played a decisive role in the passage of the reformist 1882 constitution, participating in both the drafting committee and in parliamentary deliberations that established legislative oversight over the Khedive (Al-Rafi‘i Reference Al-Rafi‘i1949; Landau Reference Landau1953). RMC MPs made no secret of their opposition to colonial rule. Collins (Reference Collins1984, 215) writes that Muhammad Galal, an RMC MP from the 1881–82 session from al-Qis in Minya province, shouted publicly that the “Khedive has sold the country to the English.” After the occupation, he was sentenced to house arrest, a three thousand Egyptian pound fine, stripped of all rank and titles by Khedival decree, and did not reappear in parliament (Collins Reference Collins1984; Subhi Reference Subhi1947). For the LE, the British occupation in 1882 provided a way to retain some semblance of power in Egypt. As Lord Cromer notes in his book Modern Egypt (Baring Reference Baring1908, 188), the LE would have been “swept into the sea,” and Egypt would be ruled by the “Sheikh class” (RMC).

LE congruence extended to Britain’s colonial economic objectives. For the LE, the foundations of their economic congruence can be traced to the mid-nineteenth century. Two treaties with Western Powers established laissez-faire economic policies in Egypt. The first is the 1838 Anglo–Ottoman Treaty of Balta Liman that dissolved state monopolies, reduced tariffs, and guaranteed British access to Ottoman markets. This treaty became binding for Egypt after its defeat in the Ottoman–Egyptian War in 1838–41. The second treaty is the Capitulations, a set of treaties with Western Powers that gave Westerners extraterritorial rights in the Ottoman Empire, exempting them from taxation and being subject to local jurisdiction. During the First Globalization Era (1850–1914), the Khedives and other LE officials in the Egyptian government, starting with Sa’id (1854–63)—who began the construction of the Suez Canal—pursued one of the most liberal laissez-faire policies in the world that encouraged the influx of European capital and financiers into Egypt (Tignor Reference Tignor1966, 38). According to Tignor (Reference Tignor1966, 42), the LE were “more responsive to the economic incentives of the modernizing market system” and more able to use their landholdings as capital to invest in modern agricultural methods, such as steam pump irrigation (Owen Reference Owen1969). By the 1860s, Egypt was the sixth largest provider of cotton for the British market (Owen Reference Owen1969, 82–3), with Britain importing 80% of Egyptian cotton. On the eve of colonial occupation, the LE held a significant share of cotton output,Footnote ¹³ were fully integrated into Egypt’s European-dominated, export-oriented market economy, and themselves benefited from lower taxation designed to support maximal cotton export to Europe. In sum, these precolonial developments led to a natural convergence in economic interests of the LE and the British occupation.

Conversely, the RMC was staunchly protectionist. The RMC opposed to European involvement in agriculture and cotton production (Baer Reference Baer1969). RMC cotton production was heavily dependent on slave labor, and the Khedive abolished slavery under the Anglo–Egyptian Slave Trade Convention of 1877. Although the emancipation of slaves was not applied immediately, this constituted a direct conflict of economic interests between the RMC and the British. In addition, RMC MPs promoted a protectionist trade policy, taxation of European capital, and supporting domestic industrialization (Dar al-Watha’iq al-Qawmiya 2017). Non-MP RMC also sent petitions to the precolonial parliament voicing opposition to European investors disrupting local production (Dar al-Watha’iq al-Qawmiya 2017). Despite the fact that the RMC were significant cotton producers, these protectionist preferences stood in direct opposition to colonial economic objectives in Egypt.

Our first empirical analysis establishes a link between colonial exposure (cotton) and redistribution of power away from the RMC to the LE in parliament. Figure 3 illustrates the aggregate evolution of the social class composition of MPs from 1824 to 1923. From 1824 to 1882, the parliament was dominated by the RMC. MPs during the precolonial period were mostly village headmen (‘umda or sheikh al-balad), mostly with sheikh or effendi titles, and predominantly from rural provinces (see Figures A2.2–A2.4 in the Supplementary Material). Following the 1882 occupation, we observe a substantial shift away from the RMC toward the LE. The share of the LE in parliament continued to rise during the colonial period, becoming the majority in the 1913–23 parliament.

Our empirical specification is a difference-in-differences (DID) model, with a continuous treatment (colonial exposure) and universal timing of the treatment across provinces (British occupation). It compares treated groups (higher cotton productivity provinces) and control groups (lower cotton productivity provinces) before and after the universal treatment (the 1882 British occupation). We estimate the following Ordinary Least Squares (OLS) regression at the MP-session level in 1824–1923:

⁽¹⁾ $$ \begin{array}{rl}{y}_{mps}=\beta cotto{n}_p\times post188{2}_s+{X}_{ps}\theta +{\alpha}_p+{\gamma}_s+{\epsilon}_{mps},& \end{array} $$

where $ {y}_{mps} $ is the social class of MP m located in province p in session s. Our outcome variables are three dummy variables indicating the LE, the RMC, and the urban middle class. The variable $ cotto{n}_p $ is the cotton yield per feddan in province p in 1877, $ post188{2}_s $ is a dummy variable that takes the value of one if session s is after the 1882 occupation, $ {\alpha}_p $ is a full set of province fixed effects that capture the cross-province baseline heterogeneity in the social class composition of parliament, and $ {\gamma}_s $ is a full set of session fixed effects that capture aggregate time shocks to the social class composition of parliament that may have affected all provinces (e.g., issuance of a new election law). The vector $ {X}_{ps} $ includes as a control variable the interaction of the post-1882 dummy variable with the cereals yield per feddan in province p in 1877. Standard errors are clustered at the province level (18 provinces).

The coefficient $ \beta $ measures the difference across higher and lower cotton productivity provinces in the evolution of the social composition of parliament before and after the 1882 occupation. We expect that the political power during the colonial period will shift away from the RMC and toward the LE. We also expect that this shift will be greater in higher cotton productivity areas than in lower cotton productivity areas, because these areas generated higher economic surplus. Hence, we expect $ \beta $ to be positive for the share of the LE, negative for the RMC, and null for the urban middle class.

The results of estimating Equation 1 are shown in Table 1. We find that higher cotton productivity provinces had a greater increase in the share of the LE and a greater decrease in the share of the RMC MPs after the British occupation in 1882, versus lower cotton productivity provinces (columns 1–4). The effects on the shares of the LE and the RMC are both statistically significant and robust to controlling for cereals productivity in 1877. The effects are large in magnitude. In column 2, provinces at the 75th percentile of precolonial cotton yield in qintars per feddan ( $ =1.76 $ ) experienced an increase in the proportion of the LE in parliament after 1882 by 21 percentage points ( $ =1.76\times 0.12 $ ), relative to provinces at the 25th percentile ( $ =0 $ ), which is three times the proportion of the LE in 1866–82. Columns 5 and 6 show that the British occupation had a null effect on the share of the urban middle class. These findings suggest that the precolonial RMC in higher cotton productivity provinces lost their parliamentary advantage during the colonial period relative to the LE more than their counterparts in lower cotton productivity provinces.

Table 1.

The British Occupation and Social Class Composition of Parliament

Note: The sample is at the MP-session level ( $ N= $ 1,102). We dropped 136 observations that are not assigned a constituency. We further dropped 16 observations with missing social class. STATA command reghdfe dropped one singleton observation that belongs to Suez province. Standard errors clustered at the province level are in parentheses. $ *p<0.10,**p<0.05,***p<0.01 $ .

The validity of Equation 1 rests upon three assumptions (Roth et al. Reference Roth, Sant’Anna, Bilinski and Poe2023). The first is the parallel-trends assumption; higher cotton productivity provinces would have exhibited a similar trend in the evolution of the social composition of their members of parliament to that of lower cotton productivity provinces, were it not for the British occupation.Footnote ¹⁴ The second assumption is no-anticipation; higher cotton productivity provinces would not have experienced a shift in the social class composition of their MPs in the last precolonial session right before the British occupation. The third assumption is that there were no other time-varying shocks that happened in or after 1882 and that affected cotton provinces differently.Footnote ¹⁵ We provide evidence in support of the first two assumptions by examining the pre-1882 trends in the social composition of MPs by cotton productivity in 1877. To do so, we allow the effect of cotton productivity to vary by parliamentary session:

⁽²⁾ $$ \begin{array}{rl}{y}_{mps}={\displaystyle \sum_{s=1824}^{1923}}{\beta}_scotto{n}_p+{\alpha}_p+{\gamma}_s+{\epsilon}_{mps},& \end{array} $$

where $ {\beta}_s $ is estimated for each session from 1824–37 to 1913–23, with the omitted baseline session being 1881–82, the last session before the British occupation. Under the parallel-trends and no-anticipation assumptions, we would fail to reject that $ {\beta}_s=0 $ for each pre-1882 session. We present the estimated regression coefficients without and with controls for cereal production in Figures A3.5 and A3.6 in the Supplementary Material, and find support for both assumptions.Footnote ¹⁶

The third assumption is supported by Egyptian historiography. Given that the 1882 British occupation was among the most significant junctures in Egyptian modern history, other time-varying shocks (e.g., the 1883 election law, cotton expansion, Suez canal concession) were either related to, or resulted from, it.

We conducted a wide range of robustness checks that we describe in Section A4 of the Supplementary Material. First, we examine an alternative explanation for our results based on Gerring et al. (Reference Gerring, Ziblatt, Van Gorp and Arévalo2011). We find that precolonial cotton productivity retains its magnitude and statistical significance even when accounting for precolonial state capacity, distance to Cairo, and other geographic controls (Saleh Reference Saleh2013). This boosts our confidence that precolonial cotton productivity measures the degree of exposure to the British occupation, and not precolonial state capacity or geography. Second, the results are robust to alternative measures of cotton productivity—including nonlinear specifications and the Food and Agriculture Organization crop suitability index (FAO and IIASA 2012)—and to estimating the standard errors using the Wild Cluster Bootstrap that accounts for the small number of clusters (provinces). Third, the results are not driven by MPs whose constituencies, occupations, or honorific titles changed over time, for example due to upward social mobility, suggesting that our results are driven by colonial policies that affected the selection of MPs based on their social class origins. Fourth, our results are not sensitive to the way we classified MPs into social class origins. When we employ occupational and honorific titles as dependent variables—instead of our composite social class measure—we find that higher cotton productivity provinces had a greater increase in the proportions of bureaucrats, and pashas and beys—who were more likely to comprise the LE according to our definition—and a greater drop in the proportions of effendis and sheikhs, and village headmen and notables—who were more likely to comprise the RMC according to our definition. We also find larger effects when we use the session-varying social class of MPs. These larger effects suggest that there was upward class mobility for some members of RMC to the LE during the colonial period.

Our main findings indicate that higher cotton-productivity provinces had a greater increase in the proportion of LE MPs, and a greater decrease in the proportion of RMC MPs after the 1882 occupation, relative to lower cotton-productivity provinces. In the next section, we substantiate the colonial economic logic undergirding the redistribution of power toward the LE.

Mechanisms

Given the LE’s congruence with, and the RMC’s opposition to, British interests, we explore whether the redistribution of political power toward the LE that we documented in Table 1 is greater in cotton-producing provinces that had higher levels of LE congruence and/or RMC opposition during the precolonial period. To this end, we estimate the following regression model that allows the main effect to vary by the precolonial political opposition of the RMC, and the precolonial economic congruence/opposition of the LE/RMC:

(3) $$ \begin{array}{rl}\begin{array}{ll}{y}_{mps}=\qquad & {\beta}_1elit{e}_p\times cotto{n}_p\times post188{2}_s\qquad \\ {}\qquad & +\hskip0.3em {\beta}_2cotto{n}_p\times post188{2}_s\qquad \\ {}\qquad & +\hskip0.3em {\beta}_3elit{e}_p\times post188{2}_s\qquad \\ {}\qquad & +\hskip0.3em {X}_{ps}\theta +{\alpha}_p+{\gamma}_s+{\epsilon}_{mps},\qquad \end{array}& \end{array} $$

where $ elit{e}_p $ measures the precolonial political opposition of the RMC, or the economic congruence/opposition of the LE/RMC in province p. Footnote ¹⁷ While we do not observe the precolonial political congruence of the LE, the RMC’s political opposition should be interpreted relative to the LE.

We measure RMC political opposition at the province level in two ways. First, we capture individual MPs’ support for ‘Urabist ideals using their documented support for executive constraints by parliament in the precolonial parliamentary minutes. We hand-coded MP speeches from the 1866–82 parliamentary sessions (Dar al-Watha’iq al-Qawmiya 2017) as pro-democratic (and therefore pro-‘Urabist) if the substance of the speech supported formalizing executive constraints, legislative oversight, or electoral reforms to curb the power of the Khedival regime.Footnote ¹⁸ We then counted the number of pro-democratic speeches made by each MP in all precolonial parliamentary sessions. We aggregated this measure to the province level by dividing the total number of pro-democratic speeches made by MPs in each province in 1866–82 by the total number of MP-session observations in that province in 1866–82, which captures the average number of precolonial pro-democratic speeches per MP-session in the province. This measure enables us to test whether the colonial administration targeted provinces where the RMC voiced more reformist, or oppositional, views prior to the occupation.

The second measure of RMC precolonial political opposition is based on the British arrest records from the ‘Urabi Revolution. These arrests took place shortly after the 1882 occupation. We compiled the list of all arrests in the ‘Urabi Revolution across Egypt from the British National Archives (Foreign Office 1882). Data on arrests include individuals’ name, locality, and occupation, so we are able to identify the number of village headmen arrests in each province to capture the RMC support for the ‘Urabi movement outside of parliament.

We measure the precolonial economic congruence of the LE at the province level by the share of ‘ushuri agricultural land from the 1877 Statistical Yearbook. ‘Ushuri land consisted of large estates that were taxed at a reduced rate than the usufruct (kharaj) land that belonged to the peasantry (including the RMC) (Abbas and El-Dessouky Reference Abbas and El-Dessouky2011). This variable captures the precolonial capacity of the LE to produce cotton. As discussed above, large estates were more open to European capital, so this variable arguably captures the LE’s congruence with British economic interests. We measure the precolonial economic opposition of the RMC at the province level by the proportion of slaves in the population. The cotton boom in 1861–65 caused the LE to increase their coercion of local labor in order to raise their cotton production, which took more local workers out of the wage labor market. Faced with a reduced supply of wage labor, the RMC responded to the cotton boom by purchasing more slaves, in order to compete with large landowners in cotton production. We thus use the proportion of slaves in the province to capture the RMC’s capacity to produce cotton before the British occupation. We computed this variable from the 1882 (precolonial) population census—the earliest census following the abolition of slavery in 1877—which records the number of Sudanese people in each district (Ministère de l’Intérieur Reference Ministère1884).Footnote ¹⁹ Given the British role in the abolition of slavery in 1877, this variable arguably captures the RMC’s opposition to British economic interests.

Equation 3 provides a quantitative test of our elite congruence theory, in both the political (regime survival) interpretation and the strategic (in Egypt’s case, economic) interpretations. RMC precolonial political opposition measures their threat to rebel (à la Boix and Svolik Reference Boix and Svolik2013), whereas the precolonial economic congruence (or opposition) of the LE (or the RMC) captures their capacity to produce cotton, and hence their credible promise (or threat) to promote (or disrupt) colonial economic interests, given their precolonial liberal (or protectionist) economic policies. So, according to our theory, we expect $ {\beta}_1 $ —the coefficient on the triple interaction term—to be positive for the LE MP share and negative for the RMC MP share. Among higher cotton-productivity provinces, the shift toward the LE and away from the RMC should be greater in provinces with a relatively more politically oppositional RMC, more economically congruent LE, and more economically oppositional RMC, during the precolonial period.

The findings are shown in Table 2.Footnote ²⁰ Consistent with our theory, we find that the impact of the British occupation on the proportion of the LE is greater in higher cotton-productivity provinces with a higher number of pro-democratic speeches per MP-session in the precolonial parliament (column 1). Column 5 shows that the coefficient on the triple interaction term is negative for the share of the RMC as expected, but is not statistically significant. These two findings suggest that the precolonial political opposition of the RMC relative to the LE increased the shift toward the LE during the colonial period in higher cotton-productivity provinces.

Table 2.

Mechanism: Precolonial Political and Economic Congruence of Precolonial Elites

Note: The sample is at the MP-session level ( $ N= $ 1,102). We dropped 136 observations that are assigned to missing constituency. We further dropped 16 observations with missing social class. STATA command reghdfe further dropped one singleton observation that belongs to Suez province. In columns 1 and 5, seven additional observations are dropped because they belong to Rosetta, which had no MPs in 1824–82. In columns 3 and 7, the omitted quartile of the land share of large estates is the second quartile. Provinces at the first quartile all have 0 cotton productivity, and so the triple-interaction term “post-1882 × cotton × large estates land share (Q1)” cannot be separately identified. Standard errors clustered at the province level are in parentheses. $ *p<0.10,**p<0.05,***p<0.01 $ .

We fail to find evidence that the extra-parliamentary involvement of the RMC in the ‘Urabi Revolution played a role in the colonial shift toward the LE. Columns 2 and 6 show that the number of village headmen arrests during the ‘Urabi Revolution does not drive the impact of the British occupation on the proportions of the LE and RMC MPs. This suggests that the involvement of the RMC in anti-colonial mass politics outside the parliament was not a decisive factor in the shift in representation toward the LE under the British.

We find that the precolonial economic congruence of the LE, as captured by the land share of large estates, and the precolonial economic opposition of the RMC, as captured by the proportion of slaves in the province, are both important drivers of the impact of the British occupation on the MPs’ shift toward the LE (columns 3 and 4), and away from the RMC (columns 7 and 8). The coefficients on the triple interaction terms are large in magnitude and statistically significant, suggesting that the impact of the British occupation on the social composition of MPs is more substantial in higher cotton-productivity provinces where the LE were more economically congruent, and the RMC was more economically oppositional, before colonialism.

Taken together, these findings support our elite congruence theory in both its political and strategic (economic) interpretations. The political congruence of the LE enabled British colonial regime survival until the eruption of the 1919 anti-colonial revolution. Additionally, the economic congruence of the LE promoted British strategic goal of maximizing economic surplus, resulting in the British recouping Egypt’s outstanding debt by WWI.

It is important to consider, however, whether there is a counterfactual in which a politically oppositional class could ever be favored (co-opted) by the colonizer if they were strategically congruent. The British Mandate in Iraq is a useful example. The British were forced to navigate a fraught relationship with a class that politically opposed colonial occupation, but stood to benefit economically from British rule: tribal sheikhs. British Prime Minister Asquith said His Majesty’s forces occupied Mesopotamia (later Iraq) in 1914 “to safeguard our interests in the Persian Gulf [and] to protect the oil fields” (Kadhim Reference Kadhim2012, 53). Tribal sheikhs played an important role in the strategic objectives of Britain’s Mandate in Iraq by maintaining local order in exchange for preferential access to land and lower taxes. Yet, even those sheikhs who stood to gain economically from cooperation with the British participated in the 1920 anti-colonial Iraqi revolution that was brutally repressed by British military force. Instead of cutting these sheikhs out of the postrevolutionary colonial political system, the Iraq government increased sheikhaly representation in the new parliament and provided even more economic benefits in exchange for their cooperation (Kadhim Reference Kadhim2012). This suggests that strategic congruence was more decisive in shaping British power-sharing in Iraq than the threat of rebellion (political congruence).

The Colonial Tools of Reengineering the Parliament

In our final analysis, we explore how the British authorities reengineered the Egyptian parliament to redistribute power among the precolonial domestic elites. Based on a qualitative examination of colonial correspondence and a comparison of the precolonial and colonial electoral laws (see Section A8 of the Supplementary Material), we identified three principal changes that the British made to the parliament that redistributed power toward the LE: (1) reducing the number of new entrants (incumbency) by creating barriers to candidacy, (2) increasing the number of appointees who served for life, and (3) adding a second chamber to the legislature.

To examine the contribution of these policy changes to the effect of the British occupation on the social class composition of MPs, we use the same specification as in Equation 1 where we decompose each social class—the outcome variables—into sub-groups defined according to the policy in question. We first examine the MP persistence tool by classifying MPs within each social class and parliamentary session into new entrants, those who did not serve before a given session, and incumbents, those who served at least once before that session (see Figure A5.7 in the Supplementary Material). Second, we examine the dynastic persistence tool by classifying MPs within each social class and session into new entrant dynasties, those MPs who are from dynasties that did not serve before a given session, and incumbent dynasties, those MPs who are from dynasties that served at least once before that session (see Figure A5.7 in the Supplementary Material). Third, we examine the appointment tool by classifying MPs within each social class and session into appointed and elected (see Figure A5.8 in the Supplementary Material). Fourth, we examine the upper house tool by classifying MPs into those who serve in the upper house and those who serve in the lower house (see Figure A5.9 in the Supplementary Material.)

We then investigate the extent to which these reengineering tools may explain the redistribution of power toward the LE. The results are shown in Tables 3 and 4. To interpret these results, recall that the main effects on the proportions of the LE and the RMC are 0.115 and $ -0.115 $ , respectively (columns 2 and 4 of Table 1). Part (a) of Table 3 decomposes these effects into incumbent and new entrant MPs, showing that more than half of the effect is driven by incumbent MPs. Part (b) shows that almost all of the main effects are attributable to MPs from incumbent, rather than new entrant, dynasties.Footnote ²¹ This reveals that MP and dynastic persistence was indeed an effective tool in shifting parliamentary representation in favor of the LE in higher cotton-productivity provinces during the colonial period. Put differently, the colonial authorities reengineered the parliament in favor of the LE by selecting MPs and dynasties from the LE in higher cotton-productivity provinces who persisted across parliamentary sessions.

Table 3.

Colonial Tools of Social Reengineering of the Parliament: MP and Dynastic Persistence

Note: The sample is at the MP-session level. The regressions in part (b) dropped 55 MP-session observations who are without a family name (i.e., only first name recorded); $ N= $ 894. Standard errors clustered at the province level are in parentheses. $ *p<0.10,**p<0.05,***p<0.01 $ .

Table 4.

Colonial Tools of Social Reengineering of the Parliament: Appointment and the Upper House

Note: UH and LH refer to the upper and lower houses, respectively. The sample is at the MP-session level; $ N= $ 949. Standard errors clustered at the province level are in parentheses. $ *p<0.10,**p<0.05,***p<0.01 $ .

Part (a) of Table 4 disaggregates the main effects across elected and appointed MPs. It shows that most of the positive effect on the LE is primarily driven by elected MPs, and secondarily by appointed MPs, whereas the negative effect on the RMC is driven by their loss of elected MPs. When we disaggregate the main effects by the upper and lower houses in Part (b), we found that the effect on the LE is primarily driven by their representation in the lower house, and secondarily by the upper house, while the effect on the RMC is driven by their colonial penalty in both houses.

To summarize, while the three tools of reengineering the parliament—MP and dynastic persistence, the appointment mechanism, and the creation of an upper house—were all employed during the colonial period, the colonial authorities shifted the parliament in favor of the LE by facilitating the election of MPs and dynasties from that class into the lower house and by appointing LE members for life into the upper house. These MPs and dynasties were more likely to preserve their parliamentary seats throughout the colonial era. We argue that, by shifting parliamentary representation toward the LE, the British guaranteed that power was concentrated within the elite that was most congruent with colonial interests.