Leader selection and why it matters: Education and the endogeneity of favouritism in 11 African countries

• Published date: 01 August 2023
• Author: Laura Maravall,Jörg Baten,Johan Fourie
• Date: 01 August 2023
• DOI: http://doi.org/10.1111/rode.12981

REGULAR ARTICLE

Leader selection and why it matters:

Education and the endogeneity of favouritism

in 11 African countries

Laura Maravall

1

| Jörg Baten

2

| Johan Fourie

3

1

Department of Economics, Universidad

de Alcal

a, Alcala de Henares, Spain

2

Department of Economics, University of

Tübingen, Tubingen, Germany

3

LEAP, Department of Economics,

Stellenbosch University, Stellenbosch,

South Africa

Correspondence

Laura Maravall, Department of

Economics, Universidad de Alcal

a, Alcala

de Henares, Spain.

Email: laura.maravall@uah.es

Funding information

Deutsche Forschungsgemeinschaft,

Grant/Award Number: SFB 1070

Resource Cultures

Abstract

A large literature on favouritism argues that leaders

favour their own ethnicity or administrative birthplace.

We question the assumption that these leaders are exoge-

nously selected for office. Using historical censuses from

11 African countries, we show that leaders are selected

from more advanced regions. In other words, our sample

shows that African leaders were created by colonial (and

pre-colonial) institutions, which often meant large educa-

tional differences between regions. Our paper's historical

perspectiveshowsthattheseoften-overlooked institutions

can account for much of the variation in post-colonial

outcomes. Favouritism was at least partially endogenous.

KEYWORDS

Africa, birthplace, economic history, favouritism, leader

selection, leadership

JEL CLASSIFICATION

F54, I25, N00, O15

1|INTRODUCTION

The Malawian University of Science and Technology was established by an Act of Parliament

in 2012 when the then president of Malawi, Bingu wa Mutharika, secured an 80 million US dol-

lar loan from the Export–Import Bank of China. The university is situated in Thyolo, a tea-

Received: 27 September 2021 Revised: 2 November 2022 Accepted: 29 January 2023

DOI: 10.1111/rode.12981

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distrib ution and

reproduction in any medium, provided the original work is properly cited.

© 2023 The Authors. Review of Development Economics published by John Wiley & Sons Ltd.

1562 Rev Dev Econ. 2023;27:1562–1604.

wileyonlinelibrary.com/journal/rode

growing district about 40 kilometres south of the city of Blantyre and with a mean education

level per individual well below the national average during the colonial era.

1

The reason for this

strange-seeming choice of location was that it is close to Mutharika's birthplace and his farm,

Ndata, where he was buried in 2012.

This example of a particular region being favoured by a person in power is not unique and

has been extensively explored in the literature. Many studies show that favouritism is present in

many parts of the world (Alesina et al. 2016; De Luca et al. 2018; Dimico 2017; Francois

et al. 2015). Favouritism has also been shown to have many consequences, including conflict

(Arriola, 2009), aid allocation (Dreher et al., 2019; Isaksson, 2020), identity formation

(Koter, 2019) and provision of public goods such as roads (Bandyopadhyay & Green, 2019;

Ejdemyr et al., 2018).

However, although much has been written about favouritism and its long-run conse-

quences, there is less about how leaders are selected. Many authors implicitly assume in their

empirical framework that leaders are exogenously selected for office or use research methods

that allow ‘differencing out’time-invariant variables that may increase the likelihood of a

region becoming a leader's birth region.

2

There are some recent exceptions to this neglect of

leadership selection phenomena. Ricart-Huguet (2021) demonstrates that colonial investment

in education determined the unequal distribution of political power in the post-colonial period

(1960–2010) across 16 African countries. He shows that post-colonial ministers are indeed ‘a

byproduct of education-based recruitment into the colonial state’(p. 2549), thus providing evi-

dence of the selection of leaders into office.

In line with Ricart-Huguet (2021), we argue that some regions may have been more likely to

produce leaders because of certain characteristics, such as a high degree of colonial education

investment. We are the first, though, to investigate the colonial differences between the regions

where African leaders were born at an individual level. We focus on how colonial powers

selected some regions for more investment than others. To do this, we first review the various

mechanisms discussed in the literature (mainly from the 1960s) by which colonial education

increased the likelihood of the emergence of post-colonial leaders from specific regions. We

then identify the birthplace of 33 post-independence leaders and examine the differences in

colonial education levels from 1930 to 1970. Using census data from the IPUMS International

database (Minnesota Population Center, 2020), we first build a cross-sectional database where

the unit of observation is the individual. We then assess the probability of completing primary

education depending on whether individuals were born in the same region as the leader before

independence and whether they attended primary education before or after independence

(defined as from ages 6 to 13). Our results show that the individuals in the regions that were

the birthplaces of post-independence leaders of our 11 countries had a higher probability of

completing primary education, and that the difference was larger during the period before inde-

pendence. We limit our sample in various ways to confirm that our results hold.

To test the robustness of our results, we aggregate the individuals into larger geographical

units. To do this, we calculate averages of our educational variables: years of schooling, numeracy

and the Gini index, capturing the direct input of education, the output of education, and the

inequality in education. We then run repeated cross-sections for the birth decades from 1930 to

1970. The results confirm that the leaders of our 11 countries came from regions that were more

advanced in terms of colonial education. Finally, using Ricart-Huguet's (2021)method,wetest

whether the probability of becoming a leader's birth region is explained by the degree of colonial

education (measured by the early presence of missions in that region). The results show that colo-

nial education increased the probability that a leader would be born in a given region.

MARAVALL ET AL.1563

Our main contribution is to provide a historical perspective on favouritism. Our results cau-

tion against the strong causal interpretations that are common in the literature. We find that

regions of more favourable education more often produced political leaders than other regions.

In line with Ricart-Huguet (2021), we argue that leaders’regional origins influence outcomes

today, but leader characteristics themselves are not random. We demonstrate this by taking into

account varying standards of education during the colonial period and finding a strong partial

correlation between colonial education and post-colonial education outcomes. We find that the

post-colonial leaders of our 11 African countries mostly came from regions with better educa-

tion in the colonial era, violating the exogeneity requirements of earlier causal interpretations.

To be more precise about this, our findings imply that the conventional ‘difference-in-differ-

ence’approach might not be sufficient to deal with the endogeneity issue. This is because that

approach depends on the assumption that while areas might be different, they will remain on

parallel trends. While this might be true in some cases, it is unlikely to be true for social invest-

ments and infrastructure where there is strong path dependence. Wantchekon et al. (2015), for

example, show that early education advantages are quite persistent and even increase over

time.

3

What this implies is that the parallel trends assumption behind the ‘difference-in-differ-

ences’assumption might be invalid if regions that produce leaders had early advantages. Not

only are regions different, but pre-independence trends across regions are not parallel, while

post-independence trends are unlikely to be parallel.

In this study, we focus on administrative birthplace but also consider ethnicity. In the

case of Africa, most scholars consider favouritism and the provision of public goods as an

ethnic issue, related to the ethnic identity of the leader and their citizens. Few studies have

considered birthplace as a separate indicator to ethnicity. Even though African presidents,

as Posner (2005) notes, often build schools, clinics and roads in their home areas, most

scholars tie this behaviour to ethnic favouritism rather than birthplace.

4

Those scholars

who have considered birthplace separately from ethnicity have found significant effects.

Ahlerup and Isaksson (2015, p. 151) used survey questions about unfair treatment to distin-

guish between ethnic and regional favouritism. They found evidence of both and concluded

that focussing on only one type of favouritism ‘risks overlooking important nuances’.

Hodler and Raschky (2014), using satellite data on nighttime light intensity and information

about the birthplaces of 126 countries’political leaders between 1992 and 2009, found more

intense nighttime light in the birthplace of the current leader. They argue that this is evi-

dence of ‘regional favouritism’. When we do run the main empirical analysis at the ethnic

level, however, our results hold. Although we believe birthplace to be a better measure, it is

not essential for our argument regarding the selection of post-colonial leaders of our

11 countries; that is, that they are selected from regions that had better pre-colonial

education.

Our paper is relevant to two areas of the literature. It adds to the favouritism literature: our

historical perspective shows that the 33 leaders in our sample of 11 African countries were

strongly selected from regions with high colonial (and pre-colonial) education levels. Control-

ling for colonial education investments could therefore moderate some of the over-large post-

colonial favouritism findings. It also adds to the colonial education literature: we use a database

that allows us to carry out a comparative analysis of colonial education, and we also draw ren-

ewed attention to a literature on colonial education and the emergence of post-independence

leaders that is often neglected in current studies. We contribute a novel method to this debate,

too: our estimation strategy is versatile, as we run regressions at both individual and regional

levels. For the latter, we include controls for spatial autocorrelation. Our results remain

1564 MARAVALL ET AL.