Searching for Sanctuary: Government Power and the Location of Maritime Piracy

Abstract

Recent systematic work on the incidence of maritime piracy shows the importance of various political, economic, and geographic correlates at the country level. Yet these correlates tell us little about the determinants of piracy location off states’ coasts, despite the fact that piracy is well known to cluster locally. Conceptualizing pirates as strategic actors who consider the risk of detection and capture, this article argues that states’ ability to project power over distance affects pirates’ decisions on where to organize and operate. As state capacity increases, piracy will locate farther away from government power centers, whereas piracy can flourish closer to state capitals in weak states that struggle to extend control over space. Using geocoded data from the International Maritime Bureau for the 1996–2013 period, results show that increases in state capacity are associated with greater median capital--piracy distances. These findings are robust to several changes in model specification. Our results have important implications for the study of piracy and crime.

KEYWORDS:

• geography
• maritime piracy
• state capacity
• transnational crime

Notes

¹ In 2011, the International Maritime Bureau (IMB) reported that piracy off East Africa (mostly Somali) represented 37% of global pirate attacks. By 2013, Somalia piracy had fallen to about 4% of worldwide attacks. In contrast, piracy in the Gulf of Guinea represented less than 5% of worldwide attacks in 2011, but that number had increased to nearly 20% in 2013. Pirates off Indonesia represented less than 20% of the global attacks in 2011 but in 2013 were responsible for nearly 50% of the worldwide attacks.

² The average State Fragility score for countries without piracy is about 7.5. For countries that experience maritime piracy, it is nearly 70% higher at 12.7. GDP per capita similarly varies. Per capita GDP of piracy prone countries is around $5,000. It is over three times higher in countries not suffering from piracy (see Prins et al. 2014).

³ Lai (2007) also notes that transnational terrorist organizations tend to survive in weak states.

⁴ Marchione and Johnson’s (2013) work on the spatial clustering of piracy incidents is an important exception but does not focus on how institutional or economic factors affect piracy location.

⁵ It is important to note that governments are also strategic actors who are selective in choosing areas for investment and the provision of public goods. While conceptualizing states as strategic actors is not inconsistent with our claim about the strategic selection of location by pirates, the underprovision of resources by the government could also lead to grievances against the state. Pirates may thus be motivated by grievances in addition to more opportunistic goals. While a more comprehensive assessment of this mechanism is beyond the scope of this article, it would seem likely that aggrieved individuals would primarily direct their discontent against that state rather than nonstate targets.

⁶ A robustness check using an alternative data source is presented in Model 5, Table 1. Data for this model come from the Anti-Shipping Activity Messages (ASAM) data collected by the U.S. National Geospatial-Intelligence Agency. These data are similar to the incident-level information from IMB, but we have not yet had an opportunity to examine the ASAM data closely to check for duplicates, ensure precise geocoding information, or to compare the ASAM incidents to IMB data.

⁷ When GPS coordinates were missing, IMB lists the port location, the distance and direction from a point, the territorial waters, the body of water, or where along a route an incident occurred. We used ArcGIS to identify port and point coordinates. For territorial waters and bodies of water, we used centroid coordinates. For incidents en route between two points, we assigned coordinates for the halfway point. For all incidents, we note the location precision so users can drop incidents with less precise coding.

⁸ For the 1996–2013 time period, over 90% of pirate incidents recorded by the IMB occurred within the territorial waters (12 nm) and exclusive economic zones (200 nm) of states.

⁹ A robustness test using mean distances is presented in Model 7. Additionally, the main model in Table 1 includes all incidents, including attacks that occur while vessels are steaming and those that occur while ships are stationary. However, since pirates cannot find permanent sanctuary at sea, we show a robustness test that includes only incidents that occur while vessels are berthed at port (Model 3).

¹⁰ In models not shown, we also specified models using unlogged median distance measures, which did not change our results.

¹¹ The Heckman model requires specification of the exclusion restriction for identification purposes. The exclusion restriction requires the inclusion of one or more covariates that influence the incidence of piracy in the selection equation but do not independently affect the location of piracy in the outcome equation (Woolridge 2010). Finding such an instrument in the selection stage is difficult, since most correlates of piracy incidence can quite plausibly be linked to piracy location. Still, we think that measures of population size and peace years could be such covariates. Population size is an indicator of the labor force that can be recruited by pirate leaders, and evidence links larger populations to more pirate attacks (Daxecker and Prins 2013; Hastings 2009; Jablonski and Oliver 2013). Past experience with piracy has also been shown to affect the future incidence of piracy (Daxecker and Prins 2013). Importantly, we do not expect population size and peace years to impact the distance between a country’s capital city and the location of pirate incidents. While population size could indirectly affect distance because more populous countries tend to be larger (and thus have greater distances on average), our models already control for country size by including land area in the selection and outcome equation. For peace years, we cannot think of plausible links with piracy location. To confirm that possible violations of the exclusion restriction in the Heckman model do not influence our results, Model 9 in Table 1 presents results using the less restrictive two-part model (2PM) proposed by Vance and Ritter (2014).

¹² In a robustness test in Model 6, we use an alternative indicator measuring state fragility from the Center for Systemic Peace scoring countries on the effectiveness and legitimacy of economic, political, social, and security conditions. Results support our main expectation. Considering additional operationalizations (Hendrix 2010), robustness tests not included found very similar results for other measures of state capacity, including the World Bank’s rule of law, the International Country Risk Guide’s measure of political stability, the Political Constraints Index independent judiciary measure, and the military personnel variable from the Correlates of War data set.

¹³ We discuss a model without the Somalia dummy in the following.

¹⁴ The surge of piracy in the Greater Gulf of Aden, which began in 2008, eventually stretched far out into the Indian Ocean by 2011 (all the way to the Maldives in fact). Such distances are unusual and were largely confined to the Somali case during this brief period of extraordinary activity.

¹⁵ We think that this concern is less warranted for an analysis at a relatively aggregate level, since we are examining the relationship between capacity and location only at the level of the state.

¹⁶ The positioning of capitals inland, however, could well be a result of lower state capacity but is not something we can examine here.

¹⁷ Data are collected by the U.S. National Geospatial Intelligence Agency and available at http://msi.nga.mil/NGAPortal/MSI.portal?_nfpb=true&_pageLabel=msi_portal_page_65.

¹⁸ See, for example, stories in USA Today (December 20, 2012) and the Southern Times of Africa (December 20, 2013).

¹⁹ The Contact Group off the Coast of Somalia and the UNODC have focused on law enforcement, the creation of a dedicated pirate court in the Seychelles and prisons in Somalia, and the disruption of financial flows from piracy. These efforts have apparently met with some success, as Somalia’s fragility score has decreased from 24 in 2011 to 20 in 2013.

Additional information

Funding

Funding for this research was provided by the Office of Naval Research through the Minerva Initiative grant # N00014-14-1-0050.