Asymmetric Effects of Positive and Negative Commodity Price Shocks During Civil Wars

ABSTRACT

Economic shocks, including shocks to commodity export prices, have often been related to civil conflict. However, the literature usually assumes that commodity prices have linear effects. In this paper, we hypothesize that negative and positive commodity export price shocks can have asymmetric effects: while negative shocks are likely to increase conflict, positive shocks can have a smaller conflict-reducing effect or even increase conflict. We test this hypothesis in a dataset for battle deaths during ongoing civil wars. Consistent with the hypothesis, negative price shocks increase battle deaths, but positive price shocks also have a positive effect. The positive effect is concentrated and becomes significant in countries with weak pre-conflict institutions that experience positive price shocks to fossil fuels. We conclude that a combination of institutional reforms, economic diversification, and price stabilization might reduce battle deaths in fuel-dependent countries with weak institutions.

KEYWORDS:

• Commodity terms of trade
• commodity prices
• conflict
• institutions
• resource curse

JEL CLASSIFICATION:

• D74; Q02; Q34

Disclosure statement

No potential conflict of interest was reported by the authors.

Data Availability Statement

The dataset and replication code for this paper is located at https://doi.org/10.6084/m9.figshare.21191515.v1

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10242694.2023.2191535

Notes

1. Another literature studies the effect of natural resources on democracy, corruption, and other institutional quality aspects (Vicente 2010; Ross 2012; Waldner and Smith 2015; Cassidy 2019). However, in this paper, we assume that institutional quality is pre-determined and measure institutional quality before the conflict.

2. Several observations suggest that it is important to treat the onset (i.e. peace to war transitions) and the intensity of conflict conditional on the onset as separate research questions. For example, Bazzi and Blattman’s (2014, abstract) conclude that ‘Conflict onset and continuation follow different processes.’ Intuitively, if civil wars cause institutional breakdowns, they can change the way the economy responds to economic shocks. For example, during the civil wars in Afghanistan, Colombia, Iraq, Libya, Somalia, Sri Lanka, and Syria over the last 60 years, non-armed non-state groups carved out large autonomous regions or ‘parallel states’ where they collected revenues through taxation, extortion, kidnapping, smuggling, and natural resource extraction and exports. The revenues were often used to finance the war (Keen 2000; Rubin 2000; Le Billon 2001; Bannon and Collier 2003).

3. Although the conflict literature has increasingly moved to micro-level data, macroeconomic conflict studies can also have advantages (Balcells and Stanton 2021). First, it is difficult to get subnational institutional quality data and institutions might vary less within countries. Second, it is difficult to construct comprehensive commodity export and import price data for subnational units and commodity prices are at least unconditionally correlated (Erten and Ocampo 2013). Third, subnational analysis might increase the probability of general equilibrium and strategic effects, such as population displacement or combatants shifting from neighboring units. Related, our definition of civil war requires that the central government is a party, which might make the country a ‘natural’ unit of analysis.

4. Terms of trade growth should normally increase welfare as each export good can be traded for more import goods (Mendoza 1995). Lederman and Porto (2016) discuss the effect of commodity prices on household welfare based on survey data from Africa and Latin America as well as a review of the literature. They conclude that households spend large budget fractions on commodities, they often depend on commodities to earn income, and international price changes pass through to households, suggesting households are exposed to both import and export prices.

5. The reason why the United States is included in the sample in Table 2 is that Lacina and Gleditsch (2005) count the death toll in Afghanistan after the terrorist attacks on U.S. soil on 9/11/2001 as an internationalized internal conflict.

6. The population data is from the Penn Tables Version 9.1 (Feenstra, Inklaar, and Timmer 2015), supplemented with World Bank W.D.I. and U.N. Population Division.

7. Ideally, we would have liked to construct a randomized measure of intervention (since the decision to intervene may be endogenous) and obtain information about the resources and policy objectives of the intervener.