Global Energy Governance: Trade, Infrastructure, and the Diffusion of International Organizations

Abstract

Why do states choose to join and form international governmental organizations (IGOs) that regulate energy policy? In this article we make three specific contributions to the literature on international cooperation and diffusion. First, we show that countries form and join energy IGOs in response to memberships previously gained by direct competitors among oil and gas producers and consumers. Moreover, we demonstrate that energy IGOs diffuse among countries that share oil and gas pipelines. Finally, we provide evidence that the institutional design of established energy IGOs impacts the development of their membership network. To test these hypotheses, we rely on original data on oil and gas pipelines and the design of energy IGOs as well as on a newly compiled dataset that includes 152 countries and covers 38 years (1970–2007). We employ both network analysis and spatial econometrics.

Keywords:

• energy
• international organizations
• network analysis
• spatial econometrics
• trade

Acknowledgments

The authors thank Alberto Clô, Eugenia Baroncelli, Jeff Colgan, Johannes Urpelainen, three anonymous referees, and the editor for comments. The authors also thank Alexander Hatzold for excellent research assistance. Part of this project was carried out when Leonardo Baccini was a NCGG fellow at Princeton University, whose financial and professional support the author gratefully acknowledges.

Notes

¹A supplementary appendix is available at http://whatever.org. A complete replication package is available at http://wherever.org.

²One of these is Maoz (2010).

³They are reported in the online appendix (https://sites.google.com/site/leonardobaccini/publications/journals).

⁴See the online appendix for the sample and the selection criteria.

⁵For the concept of organizational fields, refer to Di Maggio and Powell 1983. For the inter-organizational networks of IGOs, see, for example, Beckfield (2008, 2010); Cao (2009); Gulati and Gargiulo (1999) Hafner-Burton, Kahler and Montgomery (2009); Maoz (2010).

⁶The list of IGOs categorized in producer IGOs and consumer IGOs is provided in the online appendix.

FIGURE 3 Energy IGOs Network: Producers vs. Consumers—2009 (color figure available online).

⁷We are using the so-called spring-embedder requiring that the intersections of lines should be as low as possible. Therefore, those nodes in the middle of the field can be interpreted as relatively more central.

⁸See Gregory L. White, “Moscow Warns on Low Oil Prices,” The Wall Street Journal, March 31, 2009, http://online.wsj.com/article/SB123843968759570601.html.

⁹Quote reported in Gilani (1978:63).

¹⁰We acknowledge that information is a relevant issue both for consumer and producer countries. However, we speculate that consumers experience more difficulties in finding timely and reliable data to reduce market uncertainties because, unlike the producers, they do not control the energy sources. For a similar argument, see Bressand (2010).

¹¹The role of the IEF as a vehicle of information and recruiter of data mainly serves the purpose of providing consumer countries with credible news of energy production and supply from producing countries, as demonstrated by the Joint Oil Data Initiative (JODI). In exchange, producer states are directly involved in the framing of international energy security policies

¹²A similar provision has been recently included by ASEAN Energy Ministers in the new ASEAN Petroleum Security Agreement (APSA). Specifically, in the case of a shortage, net oil exporters in Southeast Asia are expected to supply petroleum products to the countries in need at discounted prices. Similarly, in the event of an oversupply, net-importing countries would purchase the products from those exporting countries.

¹³See Heather Stewart, “Oil price falls after US warns Iran over threat to close Gulf supply route.” The Guardian, December 28, 2011, http://www.guardian.co.uk/business/2011/dec/28/oil-prices-iran-supply-threat.

¹⁴In line with advice contained in Ward and Gleditsch (2008), we checked whether the inclusion of spatial lags is appropriate by calculating the Moran index, using the total number of IGOs joined by each country. The result confirms that there are statistically significant spatial correlations (at 99% level) between countries.

¹⁵Survival analysis is the appropriate approach because we are dealing with right-censored data. See also Beck (2008). The study by Elkins, Guzman, and Simmons (2006) of the diffusion of bilateral investment agreements also uses a survival model. Darmofal (2009) provides an extensive analysis of the use of survival models with spatial effects.

¹⁶The results are similar if we use Gamma frailty (see the online appendix).

¹⁷We do not include IGOs that are global (that is, that include virtually all the countries in the world) and for which energy policy is only a marginal issue. For instance, we do not include the WTO, the IMF, or the G20. We also decided not to include the EU, because energy has only recently gained preeminence in the organization, while our analysis considers the provisions established at the birth of the IGO and we disregard updates. The EU adopted a common energy policy in 2005, and the Lisbon Treaty (2007) is the first treaty which contains explicit reference to energy solidarity of member states and established a shared competence in energy matters.

¹⁸Since ASEAN and OAPEC were formed before 1970, they are not included in our dependent variable. However, both ASEAN and OAPEC are included in our spatial lags (yi, t−5), that is, they trigger the formation of other energy IGOs in the five years after their creation.

¹⁹The five-year cut-off point is also consistent with the operationalization used by Egger and Larch (2008) in explaining the proliferation of PTAs. Our results are not sensitive to a five-year cut-off point threshold: we obtain similar results changing this value to three or seven years.

²⁰The word “big” refers to countries that export or import a large amount of oil and gas. It does not refer to economic size, for instance, although the two measures are obviously correlated. 23SITC 3 did not exist until 1986 and there are problems of concordance between revision 2 and revision 3 (all these problems are documented in the online appendix).

²¹Because of outliers, we use the natural logarithm of this variable in our models below. Results shown below are not sensitive to this decision.

²²We obtain similar results if we divide export and import (plugged into our connectivity matrices) by GDP. We do not opt for this operationalization in the main analysis since that is usually used as a proxy for trade openness, which is not part of our theoretical framework.

²³The data are available at http://www.cepii.fr/anglaisgraph/bdd/distances.htm.

²⁴When plotting the Nelson-Aalen cumulative hazard estimator for Cox-Snell residuals, some variability is still expected, especially in the right-hand tail. This is because of the reduced effective sample caused by prior failures and censoring (Cleves, Gutierrez, Gould, and Marchenko 2008:216).

²⁵For an empirical application of the energy regime complex in environmental policy, see Keohane and Victor (2010).