Since colonization, Indigenous peoples in Latin America have suffered from social and political marginalization. Their rights and demands have long been ignored or suppressed. When not threatened by extinction, subjugation, or exploitation, Indigenous peoples have been romanticized, portrayed as an obstacle to development and national unity, or misrepresented as part of the large rural masses. In the past decades, however, a wave of constitutional reform has led to the recognition of Indigenous peoples’ rights in the fundamental norms of most Latin American countries. Still, the implementation of constitutional rights is not automatic. Without follow-up in ordinary law, they remain empty promises. In this article, I analyze how the extent of Indigenous substantive representation has changed over the past decades and investigate the conditions under which legislators act for Indigenous peoples. Taking stock of all legislation regarding Indigenous peoples adopted in Latin America since 1979, this study introduces the new INDILEX dataset and illustrates its value by systematically analyzing the determinants for the substantive representation of Indigenous peoples in Latin America.
In the early phases before and after independence from Spain and Portugal, Indigenous communities faced the danger of eradication, exploitation, domination, and assimilation (Mahoney Reference Mahoney2010, 75, 123). Although Indigenous movements and activists have long fought for their rights, state policy long continued to be more or less hostile to Indigenous communities (Yashar Reference Yashar2006, 259–260; Albala and Natal Reference Albala and Natal2023). At the beginning of the twentieth century, politicians pursuing so-called indigenist policies glorified Indigenous peoples as the mythical origins of Latin American statehood but viewed their contemporary existence as an obstacle to development to be solved through assimilation (Earle Reference Earle2007). This misrecognition of Indigenous identities continued when proponents of the idea of mestizaje argued for the creation of national unity through the development of an increasingly homogenized, racially mixed society (Hooker Reference Hooker2005, 289). Although portrayed as the opposite of racial segregation, this ideology has had exclusionary implications as it completely denied de facto existing racism (Telles and Garcia Reference Telles and Garcia2013, 132). The extension of social rights to the rural masses in the middle of the twentieth century perpetuated the misrepresentation of Indigenous peoples when they had to assume peasant (campesino) identities to benefit from welfare programs (Yashar Reference Yashar2006, 260). This class-based interpretation of indigeneity continued to make Indigenous identity invisible and caused it to disappear within or behind other identities (Gordillo and Hirsch Reference Gordillo and Hirsch2003, 5).
Only toward the end of the twentieth century did Indigenous identities become officially recognized. In a wave of constitutional reform, Latin American countries have increasingly recognized the demands of Indigenous movements and granted group-specific rights to Indigenous communities (Van Cott Reference Van Cott2004; Aguilar et al. Reference Aguilar, Lafosse, Rojas and Steward2010). The constitutional recognition of Indigenous rights represents a paradigmatic change in dealing with Indigenous affairs in Latin America (Kapiszewski et al. Reference Kapiszewski, Levitsky and Yashar2021). Some scholars have even praised it as the “most promising and innovative areas of Latin American constitutionalism” (Gargarella Reference Gargarella2013, 179). In contrast, other experts have stated that Indigenous rights suffer from a general “implementation gap,” preventing them from bringing change (Economic and Social Council 2006).
This implementation gap might arise from insufficient secondary legislation. Constitutional reform can be considered a manifestation of higher-order change in the normative core of a policy area. Such paradigmatic transformations of fundamental values need to be made operational through “normal” legislative policymaking (Hall Reference Hall1993; Sabatier Reference Sabatier1993). Indeed, as the Special Rapporteur of the Human Rights Council on the rights of Indigenous peoples elaborates, abstract constitutional norms are more likely to be put into practice if translated into more specific policy instruments (United Nations 2017, para. 37).
However, the transposition of constitutional provisions into ordinary law is not automatic (Van Cott Reference Van Cott2006, 286). For example, Venezuela never followed up on the Indigenous rights in its 1961 Constitution (Van Cott Reference Van Cott2003, 51). Similarly, Colombia still has to produce legislation regulating constitutional provisions on Indigenous jurisdiction (Art. 246), which is specifically required by the text of the 1991 constitution (Feo Valero Reference Feo Valero2019, 400). In contrast, Chile has a small but constant legislative output dealing with Indigenous peoples, although its constitution remains silent on them. Therefore, this article asks which factors facilitate the adoption of Indigenous rights legislation.
This research objective is essentially about the question of when and why democratic majorities legislate for minorities and underprivileged groups. This issue has been intensively studied in the literature on the substantive representation of women, from which I derive many of the theoretical arguments of this article. However, with the exception of Ewig’s (Reference Ewig2022) study on bill sponsorship in Ecuador, systematic studies on the substantive representation of Indigenous peoples in Latin America are even more scarce than those on Indigenous descriptive representation (e.g., Htun Reference Htun2016; Albala and Natal Reference Albala and Natal2023). The absence of such studies can be explained by the lack of comprehensive data.
I fill this lacuna by introducing the INDILEX database: The INDILEX database provides comprehensive coverage of legislation on Indigenous peoples in all countries in Latin America that are home to politically relevant Indigenous groups and covers the period 1979–2018. Its long time frame and comprehensive geographical coverage prevent the case selection bias that prior small-n studies have suffered from (Van Cott Reference Van Cott2010, 400). The data further complements the few existing large-n studies in the field that concentrate on constitution making using descriptive and cross-sectional analyses (Aguilar et al. Reference Aguilar, Lafosse, Rojas and Steward2010; Holzinger et al. Reference Holzinger, Roos Haer, Behr and Neupert-Wentz2019; Fuentes and Fernández Reference Fuentes and Fernández2022). The empirical analyses find that leftist presidents, Indigenous civil society organizations, and democracy promote Indigenous rights legislation, whereas data on descriptive representation is too limited to make definitive conclusions. Constitutions might be seen as general mission statements, broadly anchoring ordinary legislation (J. King Reference King2013), rather than concrete instructions fulfilled by legislators.
Literature and theory
Overview of the theoretical framework
A plethora of studies has investigated descriptive and substantive representation and their relationship since these concepts were introduced to the scholarship on women’s political representation by Pitkin (Reference Pitkin1967). Descriptive or “mirror” representation means representing others in the sense of standing for them, such as Indigenous constituents being represented by Indigenous representatives in parliament. Substantive representation, in contrast, means acting for the interests of the represented, no matter the identity of the representatives. Substantive representation furthermore consists of a procedural and an outcome dimension (Kroeber Reference Kroeber2018). The procedural dimension refers to individual representatives’ behavior on behalf of the constituents, like speeches or bill sponsorship. The outcome dimension relates to the aggregate results of the political process, like the adoption of legislation in the constituents’ interest.
Many studies on the topic implicitly or explicitly link the two types of representation. Still, the literature on the representation of women and other social groups like Indigenous peoples has also found further factors that can facilitate substantive representation (e.g., Ewig Reference Ewig2022; Weldon and Htun Reference Weldon and Htun2013). As legislative output is influenced by many more aspects than the behavior of individual Indigenous representatives (e.g., Alemán and Tsebelis Reference Alemán and Tsebelis2005), I start by theorizing descriptive representation and then turn to further explanatory factors as identified in the representation literature. The factors and hypotheses derived from this literature are summarized in Figure 1. I discuss them in turn.
Figure 1. Overview of theory and hypotheses.
Descriptive representation and the critical mass
Many studies on the representation of women or minority groups focus on descriptive representation, implicitly assuming this will lead to improved substantive representation. The argument is that the shared experience as women (or ethnic commonalities or other shared features) will result in similar policy preferences and thus make representatives more likely to advocate and act for the interests of the represented (Cowell-Meyers and Langbein Reference Cowell-Meyers and Langbein2009, 492). Related to the topic of representation quotas, a large part of this literature has investigated how many representatives are needed to effectively promote substantive representation. However, this research on the “critical mass” needed for descriptive representation to translate into substantive representation has not provided definitive results (see Mansbridge Reference Mansbridge2005; Wängnerud Reference Wängnerud2009).
In a global cross-sectional study, Ruedin (Reference Ruedin2013, chap. 3) finds that the representation of ethnic groups seems generally stronger than that of women. However, in the Latin American context, Piscopo and Wylie (Reference Piscopo and Wylie2020) state that women, Indigenous peoples, and Afro-descendants are still underrepresented in most parliaments. Given the low number of Indigenous representatives in Latin American parliaments, it is questionable whether they represent a critical mass to influence legislation. Nevertheless, given the prominence of the concept in the representation literature, I hypothesize the following:
H1: The more Indigenous members of parliament, the greater is the substantive representation of Indigenous peoples in ordinary legislation.
Critical actors and allies
A series of studies have qualified the results on the descriptive–substantive link by showing that the effect of descriptive representation can even produce a backlash (Franceschet et al. Reference Franceschet, Krook and Piscopo2012). Albala and Natal (Reference Albala and Natal2023, 242) argue that the descriptive representation of Indigenous peoples in Latin America has often failed because Indigenous representatives prioritized the agendas of the parties that brought them into parliament over Indigenous interests. Therefore, a series of scholars have advocated a shift from critical mass to critical actors. They argue that the existence of policy entrepreneurs who are willing to invest in women- or minority-friendly policy change irrespective of their identity is key to explaining substantive representation (Celis and Childs Reference Celis and Childs2008; Childs and Krook Reference Childs and Lena Krook2009).
The prior literature from the Latin American context suggests that Indigenous parties and leftist presidents could be important allies in promoting Indigenous substantive representation: Indigenous parties (which could also bring non-Indigenous representatives into parliament) have substantially affected Indigenous-friendly legislation in several Latin American countries. Examples include the Movimiento al Socialismo in Bolivia and Pachakutik in Ecuador, which “have pushed various policies to reduce ethnic inequalities and discrimination and encourage greater respect for Indigenous traditions” (Madrid Reference Madrid2012, 175).
Ewig (Reference Ewig2022) systematically disentangles the effect of Indigenous representatives and Indigenous parties in Ecuador. She acknowledges that ethnic parties are not a necessary condition to advance ethnic interests. However, she argues that in contexts such as Latin America, where traditional parties have long ignored Indigenous interests, Indigenous parties are crucial to promoting Indigenous substantive representation. Her argument is that Indigenous parties provide platforms, networks, and institutional access—assets necessary to advance minority rights legislation, which individual Indigenous representatives lack. This leads to the following hypothesis:
H2: The higher the share of seats of Indigenous parties in parliament, the greater is the substantive representation of Indigenous peoples in ordinary legislation.
In the presidential political systems of Latin America, the chief executive plays a central role in the legislative process (Alemán and Tsebelis Reference Alemán and Tsebelis2005). Therefore, support for Indigenous rights by the president as a critical actor seems likely to influence legislative output on the topic. Indeed, prior research has documented that the support of elite allies like presidents has been crucial for the constitutionalization of Indigenous rights (Van Cott Reference Van Cott2003).
Hale (Reference Hale2005) has observed that mainly neoliberal governments have implemented reforms to Indigenous cultural rights. However, as their goal was to limit Indigenous recognition to small areas and contain more radical demands, substantial multiculturalist reform can instead be expected from the opposite side of the political spectrum. This argument is supported by Yashar (Reference Yashar1999) and Van Cott (2002, 49; Reference Van Cott2006), who contend that the Indigenous mobilization has served as an effective platform to protest these neoliberal regimes, leading to more substantial multiculturalist reform with the support of the leftist political bloc. Furthermore, Indigenous communities traditionally tend to be aligned with the political left, whose agenda includes promoting marginalized and poor groups in general (Birnir and Van Cott Reference Birnir and Van Cott2007). This leads to the following hypothesis:
H3: The more leftist the government, the greater is the substantive representation of Indigenous peoples in ordinary legislation.
Social mobilization and Indigenous movements
Beyond these explanations, a part of the literature has put social movements as promoters of substantive representation into the scholarly focus (Weldon Reference Weldon2002; Weldon and Htun Reference Weldon and Htun2013). Social movements are “networks of informal interaction between a plurality of individuals, groups and/or organizations, engaged in a political and/or cultural conflict, on the basis of a shared collective identity” (Diani Reference Diani1992, 3). Despite their informal nature, scholars like Van Cott (Reference Van Cott2005) and Yashar (Reference Yashar1998, Reference Yashar2005) have stressed the importance of Indigenous movements for the recognition of Indigenous rights in Latin America because they organize Indigenous identities politically and provide a platform for Indigenous demands to be channeled into the political system.
According to Luis Macas, an Indigenous leader from Ecuador, the Indigenous movement has deliberately followed strategies inside and outside formal political institutions to generate support for their policy positions that challenged the status quo (García Reference García2013, 233). To do so, movements can make use of a varied repertoire of strategies, ranging from distributing leaflets over strikes to forming lobby organizations (Taylor and Van Dyke Reference Taylor and Van Dyke2004, 267). The pertinent literature suggests that protests and civil society organizations (CSOs) have been crucial to advancing the Indigenous agenda in Latin America. In Ecuador, for example, uprisings and protest marches in the early 1990s critically contributed to making Indigenous identities and demands visible (Andolina Reference Andolina2003, 728). The disruptive power of this protest wave led to successful policies redistributing land to Indigenous protesters (Van Cott Reference Van Cott2005, 111).
Reflecting on the resurgence of protest by social movements in Latin America since the 1990s, Silva (Reference Silva2015) finds that protest campaigns can impact policy outcomes directly and indirectly: By signaling the high salience of an issue among the population, protesters can directly place their demands on the political agenda, which is the first step to influencing the political process (King et al. Reference King, Bentele and Soule2007). Often, however, the effect of protesters is more indirect and works through influencing public opinion in their favor, which in turn can alter politicians’ policy positions (Ciccia and Guzmán-Concha Reference Ciccia and Guzmán-Concha2023, 302). With a few exceptions, Indigenous movements have renounced violent means to achieve their goals (Yashar Reference Yashar1999, 77). In contrast to violent movements, such nonviolent protest campaigns can be especially successful in reaching their goals because they generate legitimacy for their cause among the wider public (Stephan and Chenoweth Reference Stephan and Chenoweth2008).Footnote 1 This leads to the following hypothesis:
H4: The more Indigenous movements engage in protest actions, the greater is the substantive representation of Indigenous peoples in ordinary legislation.
As a second channel to influence legislation and achieve substantive representation, social movements can establish formal CSOs to lobby for their cause. Vogt (Reference Vogt2016) found that well-organized ethnic movements in Latin America were more likely to achieve descriptive representation because they provide resources and aggregate interests that allow effective lobbying. Such “organizations build political unity, achieve greater cohesion and community support, and ultimately strengthen their overall political position,” as Paredes and Došek (Reference Paredes and Došek2020, 124) found in their study on Indigenous representation in Peru. In contrast to the disruptive power of individual protest events, CSOs serve as stable platforms to channel demands and ideas directly into the political system through networking with key decision-makers or providing expertise to the public sector. Indeed, “by the early 1990s, Indigenous peoples’ organisations in the region had matured to a level at which they could influence and—sometimes—actively participate in the constitutional reform process set in motion by elites” (Van Cott Reference Van Cott2002, 45). In Bolivia, for example, Indigenous umbrella organizations have been providing draft legislation and policy proposals to legislators since the 1980s (Yashar Reference Yashar2005, 215–216). As a result of direct connections to and staff overlap of Indigenous CSOs and the ministerial bureaucracy, these proposals successfully became legislation (Rossi Reference Rossi2018, 48–49). As Indigenous movements compete with other sectors of society for access to legislators, I theorize that a higher number of Indigenous CSOs relative to the population size increases their impact on legislation. This leads to the following hypothesis:
H5: The higher the density of Indigenous CSOs, the greater is the substantive representation of Indigenous peoples in ordinary legislation.
Institutions and favorable context factors
Politicians and social movements alike are constrained by their institutional environment. Accordingly, many studies on substantive representation acknowledge factors such as democracy as important predictors (e.g., Schulenberg Reference Schulenberg2012; Koopmans and Michalowski Reference Koopmans and Michalowski2017). Beyond holding elections, democratic societies uphold intrinsic normative values such as inclusiveness, compromising, and the balancing of diversity in society (Holzinger et al. Reference Holzinger, Roos Haer, Behr and Neupert-Wentz2019, 13). Additionally, central characteristics of democracies include multiple decision-makers and multiple access points to the political decision-making process (Andrews and Montinola Reference Andrews and Montinola2004). These filter mechanisms of representative democratic systems ensure the protection of minority rights against the potential tyranny of the majority (Gamble Reference Gamble1997).
The third wave of democratization in Latin America coincided with the constitutionalization of Indigenous rights in Latin America and increased the connections between citizens and government through innovative participatory governance structures (Wampler Reference Wampler2012). Thus, democratization opens windows of opportunity for different groups in society, including Indigenous peoples, to channel their interests into legislative output (Kapiszewski et al. Reference Kapiszewski, Levitsky and Yashar2021, 22-23). Overall, democracy should facilitate the substantive representation of minorities. This leads to the following hypothesis:
H6: The more democratic a country, the greater is the substantive representation of Indigenous peoples in ordinary legislation.
Finally, constitutions and their mandate to protect minorities represent a crucial factor influencing the opportunities for Indigenous-friendly reforms. However, as mentioned earlier, there is no automatism leading from constitutional rights to their transposition in ordinary law. Although constitutions might merely pay lip service (Law and Versteeg Reference Law and Versteeg2013), it seems that the stronger the constitutional commitment to protect minority rights, the more likely they are to be respected in the legislative process. First of all, Indigenous groups might sue for their constitutional rights, as Sieder (Reference Sieder2007) shows in the example of Guatemala. Second, rational policymakers should avoid the adoption of contradicting policies.
In reality, studies outside the area of Indigenous rights show variance in the compliance of countries with their constitutions (Law and Versteeg Reference Law and Versteeg2013). This variance can range from ignorance of constitutional norms in ordinary law to the meticulous transposition of specific prescriptions in different policy areas. Understanding constitutions as mission statements (King Reference King2013), general constitutional recognition of Indigenous rights associated with overall increased legislative activity on the topic would represent some middle ground in this continuum. Generally, in a framework of legal hierarchy, in democracies, ordinary legislation can be expected to mirror the guidelines and mandates of constitutional law (Olechowski Reference Olechowski2018). This leads to the following hypothesis:
H7: The stronger the constitutional recognition of Indigenous rights, the greater is the substantive representation of Indigenous peoples in ordinary legislation.
Data and methods
INDILEX database and construction of dependent variable
As others have stated, the interests of minority groups might be hard to define a priori (Kroeber Reference Kroeber2018). Therefore, I use a wide operationalization of substantive representation, capturing all legislation adopted that addresses Indigenous peoples and their rights. For this purpose, I have collected the INDILEX dataset. Comprehensively, it covers all legislation dealing with Indigenous peoples adopted between 1979 and 2018; due to the availability of data on the independent variables, however, the analyses here start in 1989. It includes the sixteen Latin American countries hosting politically relevant Indigenous populations according to the Ethnic Power Relations (EPR) dataset (Vogt et al. Reference Vogt, Nils-Christian Bormann, Cederman, Hunziker and Girardin2015). The scope of the dataset is all laws and decrees with legislative force and national jurisdiction that mention Indigenous peoples at least once. I identified the relevant legal acts through a keyword search in legislative databases and legal gazettes. These keywords were specified a priori and included the following terms: indígena/s, nativo/a/s, originario/a/s, aborigen/es, autóctono/a/s, and indio/s, as well as the proper names of Indigenous groups as identified by EPR. The resulting collection of legal acts was then cross-checked with the Indigenous Legislation Databank from the Inter-American Development Bank and other pertinent country-specific compilations and compendia on Indigenous legislation (for a precise list of sources, see Table A1).Footnote 2
The completeness of the dataset depends on the quality of the underlying sources and the selection of keywords. Regarding the first, their official character (mainly national legal databases and gazettes) makes them very reliable sources. However, four countries exhibit systematic undersampling in earlier years due to the digital unavailability of sources and must be excluded from the analysis in the respective periods.Footnote 3
There is considerable variance in how often Indigenous issues appear in the legal acts in the sample. Considering articles as the relevant meaningful entities,Footnote 4 I refined the corpus of legal text to contain only those articles that mention Indigenous peoples. On the one hand, this procedure increases the sampling error by losing some articles that do not mention Indigenous peoples but still apply to them. On the other, it significantly reduces the workload for subsequent human coding.
As described so far, the data compilation relied only on the occurrence of the keywords mentioned already, ignoring the actual content of the legal acts. However, some of the keywords can mean something different from Indigenous. Therefore, a keyword in context analysis verified if the keywords are used synonymously to Indigenous. Subsequently, I excluded expressions like especies nativas (native species) from the sample.
Prepared like that, the data captures legislation mentioning Indigenous peoples. However, it is unclear whether the legal acts in the sample were really meant to foster Indigenous rights, as the law might just mention Indigenous peoples symbolically or even restrict them. Considering “symbolic” legislation, I follow Van Cott (Reference Van Cott2000, 43), who argues that even such provisions represent “an astounding development, [because prior] to the incorporation of these sentiments, official rhetoric proclaimed the homogeneous nature of Latin American societies.” This stance continued even after the social inclusion of workers and peasants in the mid-twentieth century and left Indigenous peoples struggling for the recognition of their ethnic identities (Rossi Reference Rossi2018, 24–25).Footnote 5 Just the act of mentioning Indigenous peoples or calling a group by their proper name officially represents them and entails an acknowledgment of their existence, which is the first step toward the recognition that Indigenous communities strive for (Taylor Reference Taylor1994; Hooker Reference Hooker2005). After decades of silencing their identities, Indigenous peoples are represented in the political discourse, not only through their demands or by their advocates but by the supreme decision-makers of the country.
Nevertheless, it should be determined that the adopted laws are not repressive or discriminatory. Therefore, human coders parsed the complete sample of articles for overtly restrictive legislation, such as taking away Indigenous rights, treating Indigenous peoples worse than others, or prescribing assimilation.Footnote 6 Each article was coded independently by two different coders.Footnote 7 After a thorough training phase, they reached substantial intercoder reliability, i.e., on average a Krippendorff’s α = 0.72 (Krippendorff Reference Krippendorff2018; Landis and Koch Reference Landis and Koch1977). I resolved coder disagreements through executive decisions after consulting the codebook (see Appendix A). As expected, only a fraction (less than 0.3 percent) of articles fell into the restrictive category. These provisions mainly refer to restrictions on traditional Indigenous leaders to run for public office. I excluded such cases from the sample. In this sense, the resulting database represents the way Latin American legislators acted for Indigenous peoples in the past four decades.
Finally, the dependent variable—adoption of Indigenous legislation—is operationalized by counting the number of articles compiled as described, adopted per country per year. Obviously, Indigenous movements push neither for the adoption of individual articles nor for the adoption of a certain number of laws (which would be the alternative operationalization). Instead, they fight for the adoption of provisions in their interest. I argue that the number of articles provides a better indicator for the number of provisions on Indigenous issues than the number of laws. For example, Venezuela adopted the Ley Orgánica de Pueblos y Comunidades Indígenas in 2005, which recognizes Indigenous rights in more than 150 articles, including policy areas like lands, political rights, and education. A year later, Nicaragua adopted the Ley General de Educación, which regulates Indigenous education in three articles. Counting the number of laws would accord both legal acts the same weight in the analysis, although the Venezuelan law recognizes Indigenous rights much more extensively than the Nicaraguan one. Therefore, I count the number of articles (and use the number of laws in robustness checks).
First glance at the data
Overall, 1,464 legal acts that mention Indigenous peoples at least once have been adopted since 1989. These laws and decrees contain a total of 5,295 articles that specifically deal with Indigenous peoples. During the 1980s, only 102 laws with 234 articles on Indigenous affairs were adopted. On average, each country has a legislative output of twelve articles regulating Indigenous rights per year (for other summary statistics, see also Table B2).
Around 2.5 percent of all legal acts adopted since 1989 contain at least one reference to Indigenous peoples, which shows that Indigenous affairs are indeed on the agenda of Latin American legislators. This circumstance could support the claim that the states are indeed working on a “new social contract” with the marginalized groups in Latin America (Silva and Rossi Reference Silva and Rossi2018, 12). However, comparing the adoption rates of Indigenous legislation in the sixteen countries, there is considerable variation in and around the general trend (see the solid gray and black lines in Figure 2). While some countries tend to stagnate over time (countries of the Southern Cone such as Argentina and Chile, as well as El Salvador, each with individual outliers), Venezuela is the only country with a markedly downward trend. However, this could be an artifact of a spike at the beginning of its shorter time frame in the sample.Footnote 8 The linear trend for the other countries is positive, and the highest values in most countries occur after 2000.
Figure 2. Number of articles mentioning Indigenous peoples per country and year (solid lines) and constitutional protection of Indigenous rights (dashed line).
As the empirical analysis focuses on within-country developments, Figure 2 presents each country on its own scale. It shows individual changes in legislative attention to Indigenous issues in each country in detail (for a depiction in standardized scales, see Figure B3). The between-country variations become apparent when considering the y-axis labels: Overall, most countries exhibit relatively low levels of Indigenous articles adopted per year. However, even at this low level, a rising trend becomes clear in most countries. The highest values occur in the Andean states (Bolivia, Colombia, Ecuador, Peru, Venezuela) and Mexico. These between-country differences might relate to the relative size of the Indigenous population, as Bolivia, Ecuador, Mexico, and Peru belong to the countries with the largest Indigenous communities (see Figure B3). However, as the share of the Indigenous population tends to be reasonably constant over time, it cannot explain temporal variation in the legislative outcomes. As indicated later here, I therefore include the demographic size of the Indigenous populations only as a control in some robustness checks.
Independent variables
The independent variables are operationalized as follows: To measure Indigenous descriptive representation, I compiled data on the percentage share of Indigenous members of parliament (MPs). So far, there is no comprehensive dataset on the ethnic composition of parliaments in Latin America (on the difficulty of collecting such data, see Htun Reference Htun2016, 34). Therefore, I relied on a variety of primary and secondary sources to identify the number of Indigenous representatives in parliament (considering only the lower houses in bicameral systems)—irrespective of their party affiliation. The data contains information on at least two legislative periods in fourteen countries. However, as the overview in Table B1 shows, the compilation has temporal and geographical gaps. This coverage is hardly generalizable, mainly covering the Andean countries, Guatemala, and Mexico, which tend to have the largest Indigenous populations. Nevertheless, it is the most comprehensive data collection to date and will allow us to get a first glance at the relation between descriptive and substantive representation of Indigenous peoples.
I operationalize Indigenous parties by calculating the seat share of those ethnic political parties that are linked to Indigenous groups, as identified by Vogt (Reference Vogt2016).Footnote 9 And I operationalize presidential ideology with data provided by Baker and Greene (Reference Baker and Greene2011; data updated by Baker and Greene until 2019). They provide estimates for each president on a general twenty-step left-right scale. On this scale, 1 represents the extreme left and 20 the extreme right. This dataset can be criticized for the variety of sources underlying its measurement of presidential ideology. However, its comprehensive coverage and fine-grained scaling make it hard to replace.
To operationalize Indigenous protests, I use the Mass Mobilization Protest (MMP) dataset by Clark and Regan (Reference Clark and Regan2016; data updated by Clark and Regan until 2018), which counts public protest actions with at least fifty participants. A protest event can be classified as Indigenous if Indigenous peoples are visibly part of the protesters (in contrast to individual Indigenous persons “randomly” participating). Following the MMP’s logic of data collection from news sources, I assume that Indigenous peoples formed a substantial part of the protesters if mentioned explicitly in the underlying news reports. Therefore, I rely on the description of protester identities in the dataset if Indigenous peoples are mentioned in generic terms or by their specific name (e.g., Ngöbe Buglé, Ch’ortí). A society’s overall tendency to engage in this type of political behavior could drive up the number of Indigenous protest events per year. Therefore, I account for the different levels of overall protest activity by expressing this variable as the share of Indigenous protest actions of all protest events.
To measure the density of Indigenous CSOs, I rely on Vogt’s (Reference Vogt2016; updated by Vogt until 2017) data on the strength of ethnic CSOs in Latin America. Using the subset of Indigenous groups in his dataset, I calculate the density of Indigenous Organizations as the number of Indigenous CSOs per hundred thousand inhabitants (population data by Feenstra et al. Reference Feenstra, Inklaar and Timmer2015).
To reflect the participatory innovations of Latin America’s democratization process, democracy is operationalized using the participatory democracy index from V-Dem (Coppedge et al. Reference Coppedge, Gerring, Knutsen, Lindberg, Jan Teorell and Bernhard2020). Considering prior research that different democracy indicators can behave quite differently (Vaccaro Reference Vaccaro2021), I use further indexes by V-Dem, Polity (Marshall and Gurr Reference Marshall and Gurr2020), and Freedom House (2021) as robustness checks.
To operationalize the constitutional mandate to protect Indigenous rights, I rely on the TradGov Constitutions dataset (Holzinger et al. Reference Holzinger, Roos Haer, Behr and Neupert-Wentz2019). I have extended the coding of those variables on the constitutional recognition of Indigenous group rights to match the observation period. Thus, I re-create that indicator on constitutional respect for Indigenous rights in a time-sensitive fashion incorporating constitutional changes. The variable ranges from 0 to 9 and indicates the number of policy areas in which Indigenous peoples enjoy constitutional rights (like culture or land). Thus, I use it as a comparative measure for the breadth of constitutional protection of Indigenous rights.
The visualization of this indicator by the dashed line in Figure 2 illustrates the wave of constitutional reform in Latin America that expanded the rights base of Indigenous peoples. It shows how early adopters like Panama and Peru preceded the veritable cascade of constitutional reform in the early 1990s, with countries like Bolivia and Mexico amending their constitutions several times to broaden the recognition for Indigenous rights. Upon visual inspection, there seems to be a clear connection between constitutional reform and spikes in Indigenous legislation like in Mexico and El Salvador. However, in countries like Colombia and Nicaragua, the ordinary legislation trails behind the constitutional reform.
Control variables
While this study focuses on domestic politics, international dynamics could represent an alternative explanation and have a confounding effect. For example, the adoption of international norms like the Indigenous and Tribal Peoples Convention (No. 169) by the International Labour Organization (ILO) could empower domestic Indigenous movements and encourage domestic Indigenous legislation. Therefore, I include a dummy variable indicating whether a country has already ratified ILO Convention No. 169 in any given year.
Although prior statistical findings are inconclusive, economic development might be related to several of the explanatory variables and the adoption of Indigenous legislation (Van Cott Reference Van Cott2006, 294; Holzinger et al. Reference Holzinger, Roos Haer, Behr and Neupert-Wentz2019). Therefore, I include the log of GDP per capita as a second control variable (World Bank 2020).
Furthermore, rising numbers of Indigenous policies could result from overall increased parliamentary activity due to factors unrelated to Indigenous politics. Therefore, I control for the overall legislative output, which I operationalize by the total number of legal acts passed per country per year. The information for this variable comes from the same sources as the data on Indigenous legislation. Given the duration of legislative processes, I include all explanatory variables with a lag of one year.
Reverse causality will most likely not be an issue for the variables on political actors and institutions. However, the anticipation of future legislation could encourage Indigenous movements to become more active in the present. In the absence of an indicator that could capture such a signaling effect, I control for past legislative output that could shape expectations about legislative activity by including the lagged dependent variable and, alternatively, its three-year moving average. In general, the independent variables are only weakly correlated (see correlation matrix in Table B4).
Statistical models
Using this data, I conduct a time-series cross-sectional analysis with country-years as units of analysis. This analysis aims to find out how the different explanatory variables influence the likelihood for the adoption of Indigenous legislation in Latin American countries. As the dependent variable is an overdispersed count variable, negative binomial regression is the appropriate analysis method (Cameron and Trivedi Reference Cameron and Trivedi2013, 89).
Due to the longitudinal structure of the data, observations within countries are likely not to be independent. As suggested by Allison and Waterman (Reference Allison and Waterman2002), I use an unconditional negative binomial model with standard errors clustered at the country level. I furthermore include country fixed effects, letting the models ignore variation between countries and only analyze temporal variation within countries. In this way, there are no confounding effects of time-constant country-specific factors like history or political culture. Although between-country comparisons might be interesting per se, the theoretical arguments suggest focusing on within-country temporal comparisons (Bell et al. Reference Bell, Fairbrother and Jones2019, 1069): How are different levels of these explanatory factors related to the adoption of Indigenous policy in a given country? This question is also most interesting from a perspective of policy implications. In contrast to static background variables, actors engaged in the policymaking process can manipulate the theorized time-variant variables, at least to a certain extent. To deal with temporal effects that affect all countries equally, I include year fixed effects.
In the robustness checks, I estimate hybrid within-between random effect regression models as an alternative that combines the advantages of both fixed- and random-effects approaches (Allison Reference Allison2009, 23).Footnote 10 Most importantly, they allow for differentiating the effects of variables within and between groups of observations (countries) in the dataset (Bell et al. Reference Bell, Fairbrother and Jones2019). There, I also include the demographic size of the Indigenous population as a control variable that might be of interest, which, however, remains rather constant over time and varies between countries only. The population shares of Indigenous groups are provided by the Composition of Religious and Ethnic Groups (CREG) Project (Nardulli et al. Reference Nardulli, Wong, Singh, Peyton and Bajjalieh2012).Footnote 11
Analysis
The results of the negative binomial regression analysis are reported in Table 1. The first model includes all hypothesized variables and seems to support the arguments on descriptive representation and on Indigenous movements’ strategies: Indigenous legislators, Indigenous protest, and Indigenous CSOs show a significant positive association with the adoption of Indigenous legislation. In contrast, neither the variable on Indigenous parties nor that on leftist presidents is significant at the 95 percent confidence level. Among the institutional variables, only the constitutional recognition of Indigenous rights is a significant predictor of ordinary legislation. However, the inclusion of the variable on Indigenous descriptive representation severely reduces and biases the sample, making the findings hard to generalize. Therefore, I run the same model again but exclude this variable. With the more comprehensive sample, Indigenous CSOs and the constitutional rights variable remain significantly and positively associated with the adoption of Indigenous legislation. However, whereas Indigenous protest loses significance, leftist presidential ideology reaches conventional levels of statistical significance.
Table 1. Correlates of Indigenous legislation
Negative binomial regression models with country and year fixed effects, robust standard errors (in parentheses) clustered on country level.
*p < 0.05. **p < 0.01. ***p < 0.001.
So far, Indigenous organizations, leftist presidents, and constitutional law emerge as drivers of Indigenous rights legislation. To corroborate these results, I conduct a series of robustness checks. First, I use different operationalizations for the core and control variables. Second, I use broader conceptualizations of the dependent variable, exclude nondemocratic regimes from the sample, shift the focus from individual articles to the number of complete laws mentioning Indigenous peoples, and estimate Poisson and hybrid models. Third, as the missing data from Ecuador, Mexico, and Venezuela could create a bias, and as the outliers in Bolivia and Colombia could drive the results, I replicate the analysis on samples with different geographic and temporal scopes. Since the indicator on Indigenous MPs suffers from missing data and drastically reduces the sample size, I exclude this variable in the additional analyses.
The results of the robustness checks can be found in Tables B5–B8. They broadly confirm these results but provide the following new insights: First, leftist presidential ideology is a consistent predictor for the adoption of Indigenous legislation, whereas Indigenous protest and Indigenous parties are not. Second, when focusing on the 2000s to 2010s, the variable on Indigenous CSOs loses statistical significance, whereas democracy becomes a significant predictor of Indigenous legislation—which, however, depends on operationalization choice. Third, the hybrid models reveal that the very minor) changes in Indigenous population size over time are not significantly associated with Indigenous legislation. However, the between-country comparison reveals that countries with larger Indigenous populations on average have adopted more Indigenous rights legislation. Finally, the constitutional variable—the factor initially motivating this research—in some models just drops under the 95 percent threshold of statistical significance but still remains one of the most consistent predictors for Indigenous legislation. Additional analyses, however, reveal that there is no consistent relationship between the constitutional mandate and ordinary legislation across specific policy areas (see Appendix C). Constitutional acknowledgment of Indigenous rights might trigger the adoption of Indigenous legislation in a broad sense. However, legislators do not directly transpose Indigenous rights from constitutional into ordinary law.
It could be questioned whether the inclusion of constitutional rights in the regression is an endogenous variable measuring the same concept as Indigenous legislation. One could indeed assume that the same actors in parliament might vote upon constitutional rights and Indigenous legislation. However, constitutional reforms occur much less frequently than the adoption of ordinary legislation and require special majorities. Furthermore, the constitutionalization of Indigenous rights has usually been undertaken in special constituent assemblies as part of more fundamental state reform (Silva and Rossi Reference Silva and Rossi2018, 16). Thus, the considerations and processes leading to constitutional change and the adoption of ordinary law are different.
Discussion and conclusion
In introducing the new INDILEX database, this article has investigated the determinants for the substantive representation of Indigenous peoples in Latin American legislation. Based on the political representation literature, I contrasted descriptive representation with Indigenous parties and political allies, social movements, and a favorable institutional environment as explanatory factors for the substantive representation of Indigenous peoples.
The empirical analysis relied on the INDILEX dataset and confirmed that Indigenous affairs are increasingly on the agenda of Latin American legislators. Bolivia, Colombia, and Venezuela are champions of Indigenous legislation in quantitative terms. Most other countries exhibit rising trends, albeit at a lower level.
The analysis suggests that descriptive representation through Indigenous MPs might have a positive association with substantive representation. However, given the gaps in the data, this finding must be treated very cautiously. Adding to the findings by Ewig (Reference Ewig2022), there was no significant association between Indigenous parties and the adoption of Indigenous rights legislation. This could mean that Indigenous parties might be able to influence the legislative agenda through bill proposals, as Ewig finds, but might be less effective in influencing how parliament votes on those bills. This suggests further disentangling of the determinants for the procedural and outcome dimensions of substantive representation and calls for further research—including data collection efforts on descriptive representation. Furthermore, the results on the ideology of presidents, who were identified as critical actors, support the prior research indicating that leftist governments tend to promote minority rights agendas.
The analysis showed that the density of Indigenous CSOs has an overall positive association with the extent of Indigenous legislation—especially in the 1990s. In contrast, I found no evidence that Indigenous protests predict Indigenous legislation. This result should belittle the role of Indigenous mobilization for democratization and the adoption of Indigenous constitutional rights (Van Cott Reference Van Cott2004). In light of Silva’s (Reference Silva2015) categorization, protest might be important to put movements’ demands on the political agenda (direct effect), but their indirect effects might not be strong enough to influence legislators’ decisions. Instead, lobbying by formal organizations might have a greater chance to influence lawmaking directly.
Democracy seems to play a positive role, but the results varied across operationalization and modeling choices (see Vaccaro Reference Vaccaro2021). This shows the importance of carefully considering which aspects of democracy the indicators actually measure. Given that most countries in the region had generally high democracy scores throughout most of the observation period, this also raises the question of whether democracy might be a precondition for substantive representation of minorities, with actual levels of “democraticness” being less relevant for the degree of substantive representation of Indigenous peoples.
Finally, constitutional protection of Indigenous rights seems to facilitate the adoption of ordinary Indigenous legislation—especially since the 2000s. The counterexamples of Chile and Paraguay, however, show that a constitutional mandate is neither a necessary nor a sufficient condition for the adoption of Indigenous legislation. Furthermore, the content of Indigenous legislation does not tend to follow the policy emphasis of the constitutional norms. My interpretation of this finding is that constitutions represent general mission statements (King Reference King2013) but are not concrete instructions fulfilled by lawmakers. Thus, although the adoption rates of Indigenous legislation are rising, the implementation gap of Indigenous rights might relate to an insufficient transposition of constitutional into ordinary law—a topic that merits further research.
In summary, these analyses on the determinants of representation confirmed that some of the factors identified by the literature on Indigenous rights in Latin American constitution making also hold for ordinary legislation. They also contribute to the literature on minority representation that is moving beyond the dichotomy between descriptive and substantive representation and support the literature emphasizing social movements, especially organizations, as determinants for substantive representation. When comparing these results to the slightly diverging insights of prior research on the descriptive representation of Indigenous peoples, such as Vogt’s (Reference Vogt2016) study, it should be noted, however, that the mechanisms promoting descriptive and substantive representation might differ.
The analysis relied on the count of articles. I have argued that the mere quantity of legislation indicates how serious legislators are about the importance of Indigenous affairs. Nevertheless, it could be questioned whether all articles are equally important in substantial terms. Given the history of Indigenous rights, even symbolic provisions are an important development (Van Cott Reference Van Cott2000, 43). Still, it would be desirable to delve further into this question.
This study contributed to scholarship on Indigenous representation in legislative outcomes by covering all sixteen countries hosting politically relevant Indigenous groups in Latin America since 1989 in a large-n correlational analysis. This methodological approach allowed for establishing broad patterns—an undertaking made possible through the original data in the INDILEX database. The database creates many opportunities to investigate the determinants and consequences of Indigenous legislation in a comparative manner. Besides analyzing more domestic dynamics, further research could also exploit the INDILEX database to analyze international influences on Indigenous legislation, such as patterns of international diffusion and convergence of Indigenous policies.
When analyzing consequences, further research could systematically map the content of the laws with socio-economic outcomes to find out if legislation achieves the intended impact. Such an analysis of the policies could rely on the manually coded variables in the INDILEX database or on computational text analysis methods. It would also be possible to investigate the impact on intra- and intergroup relations: What does it mean for the social fabric of Indigenous communities when their ancestral rights are formalized? And what does it mean for majority-minority relations? Thus, the INDILEX database could help to better understand politics in multicultural or plurinational societies.
Data availability statement
Online Appendix and replication material for this article can be accessed on the Harvard Dataverse: https://doi.org/10.7910/DVN/YB5E8L.
Acknowledgments
I acknowledge funding provided by Ev. Studienwerk Villigst e.V. and by Deutsche Forschungsgemeinschaft (DFG—German Research Foundation) under Germany’s Excellence Strategy – EXC-2035/1 – 390681379. I thank Katharina Holzinger, Christina Zuber, Luis L. Schenoni, Ana López García, Nadine Segadlo, and Anselm Fliethmann for helpful comments on earlier versions of this article. I also thank the three anonymous reviwers for their very helpful and constructive feedback. I am very grateful for the invaluable research assistance of Martin Schlatte, Carlos Pintos Perreira, and Lucas-Kori Leonhard. A prior version of this article was presented at the Virtual Annual Congress of the Latin American Studies Association, May 26–29, 2021.
Open access
