Do Natural Resources Really Cause Civil Conflict? Evidence from the New Global Resources Dataset

• Date: 01 April 2022
• Author: Andrew Stravers,Joelean Hall,Michael G. Findley,James Igoe Walsh,Michael Denly
• Published date: 01 April 2022
• DOI: 10.1177/00220027211043157
• Subject Matter: Articles

2022, Vol. 66(3) 387 –412

Do Natural Resources

Really Cause Civil

Conflict? Evidence from

the New Global

Resources Dataset

Michael Denly

1

, Michael G. Findley

1

, Joelean Hall

2

,

Andrew Stravers

3

, and James Igoe Walsh

4

Abstract

Scholars have long examined the relationship between natural resources and conflict

at the country level. More recently, researchers have turned to subnational analyses,

using either individual countries or subnational data for a small number of resources

in sub-Saharan Africa. We introduce a new sub-national dataset of 197 resources

that adds many resource types, locations, and countries from Africa, the Middle East,

Asia, Latin America, and Europe. To demonstrate the value of the new dataset,

we examine how conflict incidence varies with the value of the collective set of

resources in a given location using world prices. We then introduce new country-

specific price data, which are more relevant for conflict dynamics. Since country-

specific prices can be endogenous to conflict, we instrument country-specific prices

using U.S. and world prices. We find that sub-national resource wealth is associated

with higher levels of conflict using some specifications, though the results vary widely

by data source and world region. Using an instrumental variables strategy lends the

strongest support to this positive relationship, but only for African countries.

1

University of Texas at Austin, Austin, TX, USA

2

Technical University of Munich, Munich, Germany

3

Clements Center for National Security, University of Texas at Austin, Austin, TX, USA

4

University of North Carolina at Charlotte, Charlotte, NC, USA

Corresponding Author:

Michael G. Findley, University of Texas at Austin, 158 W. 21st ST STOP A1800, Batts Hall 2.116, Austin,

TX 78712, USA.

Email: mikefindley@utexas.edu

Journal of Conflict Resolution

ªThe Author(s) 2021

Article reuse guidelines:

sagepub.com/journals-permissions

DOI: 10.1177/00220027211043157

journals.sagepub.com/home/jcr

Article

388 Journal of Conflict Resolution 66(3)

Notably, across all of our models, we find that resources are negatively associated

with conflict in Latin America, suggesting heterogeneity of effects worth future

exploration.

Keywords

natural resources, civil conflict, civil wars, political violence

Over the last two decades, social scientists have devoted significant scholarship to

the “resource curse”— the proposition that an abundance of non-renewable natural

resources has negative political, social, and economic consequences (e.g., van der

Ploeg 2011; Ross 2015). A large segment of existing resource curse scholarship has

focused on the links between natural resources and violent conflict (De Soysa 2002;

Fearon and Laitin 2003; Collier and Hoeffler 2004; Ross 2004b, 2006; Humphreys

2005; Cotet and Tsui 2013; Lei and Michaels 2014; Bell and Wolford 2015; Esteban,

Morelli, and Rohner 2015; Paine 2016; Menaldo 2016). To date, the role of oil

wealth in fomenting conflict at the national level has received the most scholarly

attention from the resource-conflict literature. The focus on oil at the national level is

logical: oil is the world’s most valuable commodity, data on national oil production

and reserves are readily available,

1

and some national-level studies analyzing mul-

tiple resources have found few links between countries’ resource wealth and conflict

(e.g., Bazzi and Blattman 2014).

However, much recent research on natural resources and conflict has taken a

decidedly micro turn, emphasizing that oil and other resources, such as diamonds

and gold, may promote violent conflict at the local level (Nillesen and Bulte 2014).

The reason underpinning the micro-level turn is that many conflicts are local in

nature, yielding high violence in specific regions while the rest of the country

experiences little violent contention. Accordingly, Koubi et al. (2014, 12) suggest

that “the analysis of disaggregated data that are also able to capture the location and

spatial aspects of resources clearly seems to be the most effective approach” for

advancing knowledge. Such spatial natural resources data have proved crucial for

understanding local conflict dynamics (Arag´on and Rud 2013; Dube and Vargas

2013; Ma¨hler and Pierskalla 2015; Maystadt et al. 2014), the incentives for national

leaders to tolerate conflict (Koubi et al. 2014), how resources influence secessionist

conflicts (Ross 2012; Asal et al. 2016), and how profiting from resources by

rebel groups influences conflict dynamics (Fearon 2004; Conrad et al. 2019;

Walsh et al. 2018).

Primarily due to the limitations of existing natural resource datasets, only a few

published studies have analyzed how natural resources influence violence at the

local level in multiple countries (Berman and Couttenier 2016; Berman et al.

2017; Harari and La Ferrara 2018; Christensen 2019).

2

To help researchers develop

more general conclusions on the resource-conflict nexus as well as the resource

2Journal of Conflict Resolution XX(X)