Ethnic federalism as a strategy to reduce horizontal educational inequalities and accommodate ethnic diversity in Ethiopia

• In Event: A comparative analysis of inequalities in national education systems: Case studies from Ethiopia, Jordan, Uganda, and Chile

Proposal

(With Elisabeth King and Cyrus Samii)


1. Introduction

Horizontal inequalities, inequalities between groups, have been linked to violent conflict (Stewart, 2011) and slower economic development (Alesina et al., 2016) and scholars have argued they may become so embedded and persistent that they trap people into perpetual poverty (Stewart and Langer, 2008). Research suggests that horizontal educational inequalities in particular are positively associated with an increased incidence of conflict (Alcorta et al., 2018; Østby et al., 2009). For example, Østby et al., (2009) show that intraregional inequalities in access to education in South Africa increased the likelihood of conflict between 1986-2004. Horizontal educational inequalities are also thought to be especially harmful as they serve to reinforce broader economic, political, and cultural inequalities (Brown, 2011; Omoeva et al., 2015). Despite their deleterious effects, little is known about how horizontal educational inequalities can be reduced. This paper aims to examine whether an ethnic federal regime can serve as a strategy to reduce horizontal educational inequalities.

Scholars have theorized that federalism can be used to diffuse ethnic conflict and promote equality between groups (see for example Bermeo, 2002; Braubock, 2000; Ghai, 2000; Horowitz, 1985; McGarry & O’Leary, 1993; Simeon and Conway, 2001; Stewart et al., 2008). Ethnic federalism, a federal system in which the federated regions are ethnically defined, may be particularly well suited to accommodate ethnic demands and reduce inequality by allowing ethnic groups to exercise self-government and to promote their language and culture (Selassie, 2003). Others fear, however, that federalist policies may reinforce ethnic identities and lead to secession, conflict, or disintegration of the federal state (Braubock, 2000; Lieberman & Singh, 2012, 2017).

Ethiopia, the only ethnic federation in Africa, offers a unique opportunity to study whether an ethnic federalist regime that seeks to mitigate ethnic tensions through ethnic federalism has been successful in reducing educational inequalities. When ethnic federalism was introduced in Ethiopia in 1991, the Tigrayan-led Ethiopian People’s Revolutionary Democratic Front (EPRDF) precisely argued that the new system was needed in order to appease ethnic tensions and promote ethnic equality. Despite the theoretical expectations certain scholars hold of the potential benefits of federalism and the justifications politicians have given for establishing a federal regime, there is little empirical evidence to support their claims. While numerous studies have found a link between federalism and conflict incidence or intensity (Amoretti and Bermeo, 2004; Bakke and Wibbels, 2006; Brancati, 2006; Christin and Hug, 2012; Hale, 2004; Zuber, 2011), few rigorous quantitative studies have examined the effect of federalist policies on horizontal inequalities. The recent and ongoing violent conflict in Ethiopia between the now ousted Tigray People’s Liberation Front (TPLF) and the new regime make an understanding of the effect of ethnic federalism on ethnic inequalities as well as on ethnic identification even more pressing.

In this paper, we ask:

RQ1: Did the introduction of ethnic federalism in Ethiopia reduce horizontal educational inequality among ethnic groups?

While reduced horizontal inequalities may reduce conflict, ethnic federalism may in turn reinforce ethnic identification, at the expense of national identification, and promote conflict.

To understand this possible counter-effect, we also ask:

RQ2: Did the ethnic federalist educational system, which encouraged instruction in a local language at the primary level, impact levels of national and ethnic identification?

We hypothesize that after federalist policies targeting the education sector were introduced in Ethiopia in 1994, horizontal educational inequalities between ethnic groups decreased. We expect that the decentralization of education delivery, including curriculum development, to the regional and local level, as well as instruction in local languages in primary schools reduced barriers to participation for previously marginalized groups thus reducing inequalities. We hypothesize, however, that the decentralization of education to the regional and local level also served to reinforce ethnic identification at the expense of national identification.

2. Data

To answer RQ1 and understand the impact federalist education policies had on horizontal educational inequalities in Ethiopia after 1994, we use the 2016 Ethiopia Demographic and Health Survey (DHS). The 2016 DHS surveyed a total of 75,229 household members and collected information on educational attainment, the native language of respondents, which allows us to determine their ethnic group, and age, which will allow us to use cohort analysis of students at the age of 25 before and after 1994.

To answer RQ2 we use data gathered on language of instruction across all Ethiopian states in grades 1-4, 5-6, and 7-8 (Heugh et al., 2007) and geocoded Afrobarometer data from 2013 and 2020. The data gathered reveals that different regions switch from mother tongue instruction to English at different times, with some regions beginning instruction in English in grade 5, others in grade 7, and others not making the switch at all. Afrobarometer data includes information on the level of national and ethnic identification of respondents as well as levels of interethnic trust.

3. Identification strategy

To answer RQ1, we will implement a difference in differences in reverse (DDR) approach (Kim & Lee, 2018). We consider “reduced discrimination” (which we define as increased local funding for education, local control over education delivery, and delivery of education in local languages) to be the treatment since we would like to understand the effect of reduced discrimination resulting from federalist policies on horizontal education inequalities between ethnic groups. We can consider our control group, the Amhara ethnic group, to have always been treated (we can assume the Amhara were not discriminated against prior to 1994) and our treatment group, all other ethnic groups, to have been untreated prior to 1994 and treated post-1994 (we expect that while ethnic groups were discriminated against prior to 1994, federalist policies reduced this discrimination). Assuming parallel trends in educational attainment between ethnic groups after 1994, DDR allows us to identify pre-1994 cohort-effects.

To answer RQ2, we will exploit the variation in the number of years that students are exposed to native language vs. English language instruction. We will analyze whether the level of local language instruction is associated with different levels of national identification, tolerance, and perceptions of social cohesion.

Author

• Sorana Acris, New York University