Oil, Natural Gas, and Intrastate Conflict: Does Ownership Matter?

ABSTRACT

The impact of natural resources on intrastate violence has been increasingly analyzed in the peace and conflict literature. Surprisingly, little quantitative evidence has been gathered on the effects of the resource-ownership structure on internal violence. This article uses a novel data set on oil and natural gas property rights, covering 40 countries during the period 1989–2010. The results of regression analyses employing logit models reveal that the curvilinear effect between hydrocarbon production and civil conflict onset—often found in previous studies—only applies to countries in which oil and gas is extracted by state-owned companies. The findings suggest that only state-controlled hydrocarbon production might entail peace-buying mechanisms such as specific clientelistic practices, patronage networks, welfare policies, and/or coercion. At the same time, it seems that greed and grievance are more pronounced whenever resources lie in the hands of the state. Exploring the within-country variation, further analyses reveal that divergent welfare spending patterns are likely to be one causal channel driving the relationship between resource ownership and internal violence.

KEYWORDS:

• Civil war
• conflict
• comparative
• geography

Acknowledgments

I thank three anonymous reviewers for comments and suggestions to a previous version of this paper. Constantin Ruhe, Laura Albarracin, and Maik Maerten provided outstanding research assistance.

Notes

¹ It is important to note that the state is the sole owner of natural resources such as hydrocarbons or minerals in the great majority of countries worldwide. In this article, the term ownership is used as a synonym for the mode of resource extraction. Therefore, state ownership denotes instances in which natural resources are extracted by state-owned companies.

² For an overview of the possible mechanisms driving the relationship between natural resources and civil war onset, see Humphreys (2005), Ross (2004, 2006), and Le Billon (2008).

³ While Canadian and Australian oil and gas production lies nearly entirely in private hands, governments in Bahrain, Iran, Mexico, and Saudi Arabia retain total control over hydrocarbon production. The ownership pattern in the four remaining countries is mixed.

⁴ Andersen and Ross (2012), for example, demonstrate that peace-buying mechanisms associated with hydrocarbon wealth are more likely to produce regime stability when a country’s petroleum sector is state-owned.

⁵ The authors note that accounting books are unavailable for a high number of NOCs. Also, many NOCs do not undergo transparent auditing (McPherson and MacSearraigh 2007:210).

⁶ The budget of the Iraq National Oil Company, for example, was confidential.

⁷ Ross (2004) highlights the role of so-called booty futures in order to finance rebellion in countries such as Liberia, Sierra Leone, and the Republic of Congo.

⁸ The option of expropriating private hydrocarbon corporations is often not viable, as expropriation usually comes at a high cost and entails risks (Guriev et al. 2011). Moreover, some countries may simply lack the know-how to exploit hydrocarbons through state-owned companies.

⁹ As outlined in the literature review, Thies (2010) unites the civil war and state-building literatures by employing the predictions from predatory theory. The author stresses that governments are the main beneficiaries of natural resource revenue even in the presence of rebels as revenue-seeking predators and finds that natural resources may in fact enhance the capacity of the state.

¹⁰ Data on oil and gas production come from Evaluate Energy, http://www.evaluateenergy.com/.

¹¹ A list of all countries as well as the respective mean values of their ownership variables is presented in Table A1 in the appendix.

¹² This is mostly owed to the fact that some NOCs do not publicly disclose their exact production figures.

¹³ See http://www.usgs.gov/.

¹⁴ As an example, approximately 60% of Nigeria’s total oil and gas output was of unknown provenance prior to 2004 (up to 2003, Evaluate Energy only provides production figures of privately owned companies operating within this country). According to the single reports of the US Geological Survey, the market share of Nigeria’s NOCs was around 60% during the given period. Thus, we could attribute the missing production to the state.

¹⁵ For these reasons, the sum of the correspondent shares of state and privately controlled hydrocarbon production do not add up to 100% in some countries (or add up to more than 100% in the cases of Brazil, Oman, and Venezuela).

¹⁶ While missing observations are limited to single years in Azerbaijan, Equatorial Guinea, and Libya, the remaining five mentioned countries are marked by missing information for longer time periods.

¹⁷ See http://www.energyintel.com/Pages/EIG_GroupHome.aspx.

¹⁸ When calculating the unit “barrels of oil equivalent” (boe), the output of natural gas is standardized according to the energy content of one oil barrel.

¹⁹ See Table A2 in the appendix for further information on the definitions and sources for all the variables employed.

²⁰ The period of analysis reflects the availability of the Evaluate Energy data on companies’ oil and gas production. Since the coding of countries’ oil and gas production patterns is very time intensive, and generating a data set encompassing all hydrocarbon producing states worldwide was beyond the scope of our research project, we had to limit the analysis to 40 countries by applying strict selection criteria described earlier.

²¹ Resource-wealthy countries share common characteristics that might be relevant for explaining internal violence onset but often remain unobserved, thus creating spurious relationships.

²² Only country-years in which the sum of state-owned and privately owned oil and gas figures made up for more than 60% of the officially reported total hydrocarbon production were considered.

²³ The turning point after which state hydrocarbon production reduces the risk of intrastate conflict lies at approximately 90 daily barrels of oil equivalent per capita. The following 15 countries cross this threshold at least once during the period under analysis: Algeria, Bahrain, Brunei, Iraq, Kuwait, Libya, Norway, Oman, Qatar, Russia, Saudi Arabia, Trinidad and Tobago, Turkmenistan, United Arab Emirates, and Venezuela.

²⁴ All the results from the robustness checks reported are available upon request.

²⁵ Countries with a high leverage on the results are Azerbaijan, Venezuela, Egypt, Nigeria, and Congo.

²⁶ These data were taken from Taydas and Peksen (2012).

²⁷ Growth is measured by the annual growth of GDP taken from World Development Indicators.

²⁸ Source: Penn World Tables (Heston, Summers, and Aten 2009).

²⁹ Source: Database of Political Institutions (DPI).

³⁰ Source: Database of Political Institutions (DPI).

³¹ As noted by Halaby (2004:511), the random effect estimator is strongly influenced by cross-sectional variance and underlies the assumption that unobserved heterogeneity is mean-independent from the causal variable.

³² The choice of this time period was contingent on the availability of the welfare spending data from Taydas and Peksen (2012).

³³ Regarding the control variables, per capita GDP is negatively associated with welfare spending, which is in line with previous findings (Ha 2008; Rudra and Haggard 2005). Also, higher inflation increases social expenditures, suggesting that under sociopolitical pressure state leaders overcompensate for inflation. Economic growth also seems to encourage welfare expenditures. All other control variables remain nonsignificant.

³⁴ These results are available upon request.

Additional information

Funding

Financial support from the German Research Foundation (DFG), as part of the research project Governing the Resource-Conflict Nexus, is gratefully acknowledged.