Size Matters: The Effect of the Size of Ethnic Groups on Development
| Published date | 01 June 2017 |
| DOI | http://doi.org/10.1111/obes.12145 |
| Date | 01 June 2017 |
| Author | Arcangelo Dimico |
291
©2016 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 79, 3 (2017) 0305–9049
doi: 10.1111/obes.12145
Size Matters: The Effect of the Size of Ethnic Groups
on Development
Arcangelo Dimico†
†Queen’s University Belfast, Management School, 185 Stranmillis Road, Belfast BT9 5EE,
UK (e-mail: a.dimico@qub.ac.uk)
Abstract
The share of ethnic groups is one of the most important features of African politics. It affects
civil wars, representation in government positions, distributive and allocative policies. In
this paper, we use the partition of ethnic groups as a natural experiment in order to estimate
the effect of the share of these ethnic groups on development. We show that larger groups
have an advantage in terms of development and that the partition in itself does not matter
for development. This result is explained by the fact that the partition matters only when
the resulting groups are relatively small, since their lack of political representation may
weaken support for institutions, may bias policies and the provision of ethnic/regional
public goods.
I. Introduction
In the last few years, a large body of research has tried to explain why some countries still
lag behind while other countries have enjoyed steady long-term economic development.
Among the several explanations which have been offered, the institutional hypothesis has
been the one which has dominated the economic arena. Poorly performing political and
institutional structures, together with inefficient legal and court systems, are among the
primary causes of poor development (La Porta et al., 1997, 1998; Acemoglu, Johnson and
Robinson, 2001, 2002, 2005; Acemoglu and Johnson, 2005; Glaeser et al., 2004, etc.).
With regard to political institutions, Acemoglu and Robinson (2012) distinguish between
extractive and inclusive institutions. Unlike inclusive institutions, which are conducive to
long-term development, extractive institutions are designed to serve the interest of small
elites by exploiting the rest of the population and for this reason they are highlydetrimental
to development.
Africa is one of the continents which have suffered most in terms of the persistence of
extractive institutions. Inefficient property rights, patronage politics, corruption, mistrust
and unstable democratic institutions have long been proposed as sources of poor develop-
ment in the continent. Some of these institutions are the result of the slave trade (Nunn,
JEL Classification numbers: O10
*The author is grateful for valuable comments from the editor and anonymous referees.
292 Bulletin
2008; Nunn and Wantchekon, 2011). Others derive from the legacy of colonization and the
subsequent Scramble for Africa which at times has been considered even more harmful
than colonization itself (Asiwaju, 1985; Dowden, 2008).
Loosely speaking, the Scramble for Africa consists in the arbitrary and improper bor-
der design which partitioned a significant fraction of the population belonging to existing
ethnic groups. As a result, where political borders do not always coincide with preex-
isting tribal institutions (Englebert, 2000a), large shares of the population belonging to
different ethnic groups have been forced to coexist in artificial states (Alesina, Easterly
and Matuszeski, 2011). The discontinuity between pre and postcolonial institutions has
caused illegitimacy (Englebert, 2000a), civil conflicts (Fearon and Laitin, 2003; Fearon,
2006), ethno-political mobilization (Posner, 2004a,b), and particular rather than collective
policies (Miguel, 2004). A somewhat more sophisticated interpretation of the effect of
the Scramble for Africa is provided by Posner (2004a). Focusing on the partition of the
Chewa and Tumbukas peoples between Malawi and Zambia, he argues that ‘the politi-
cal, social and cultural salience of the cleavage depends on the sizes of the group that
the cleavage defines relative to the sizes of the political and social arenas in which the
groups are located’ (Posner, 2004a, p. 543). Thus, the idea that the political salience of a
cultural cleavage results from the arbitrary imposition of boundaries in itself seems open
to doubt.
In this paper, we build on the hypothesis in Posner (2004a) in order to investigate the
idea that the partitioning of ethnic groups in Africa matters only when the share of groups
resulting from the imposition of national boundaries is small in relation to the political and
social arenas in which the groups are located. To analyse the effect of partitioned groups
(and their relative share) on development, we use data at group level from Murdock (1959,
1967) which we merge with ethno-country estimates on gross domestic product (GDP)
from satellite imagery of the light density at night from the NOAA/NGDC (National
Geophysical Data Center). We initially exploit the variation across groups to show that
the negative effect of the partition is not robust to the inclusion of additional controls.
After establishing that the effect of the partition is not significant, we exploit the variation
within groups in order to estimate the effect of the share of the group on development.
Exploiting the variation within groups allows us to control for confounding variables,given
that partitioned groups belonging to the same historical homeland are likely to start from
the same level of income per capita and, share the same culture and institutions, which
then neutralizes biases of this kind. Using this sort of matching estimator, we find a close
relationship between the size of the partitioned group and development.
The robustness of the results is tested using a series of different approaches. First, we
use a regression discontinuity approach exploiting the discontinuity at the border.Then we
reestimate same models using an IV estimator and we find that estimated effects from these
two alternative estimators (IV and RDD) are quite close and not that far from OLS results.
Finally we change the dependent variable using household data from the Afrobarometer
to check the sensitivity of the results to different measures of economic development and
we still find a significant effect of the share of partitioned groups on proxies of individual
income.
In order to shed light on some of the channels through which the share of ethnic groups
may affect development,we test a set of different theories which have been proposed by the
©2016 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial