Proﬁts from Peace: The Political Economy of Power-Sharing and Corruption

Summary. — Does power-sharing drive corruption in post-conﬂict countries? We conceptualize government elites in any post-conﬂict situation as rent-seeking agents who need to ensure the support of their key constituencies to remain in power. Power-sharing institu-tions—especially cabinet-level, executive power-sharing institutions—systematically shape these rent-seeking motives. Power-sharing cabinets create political coalitions dominated by small circles of government and rebel elites with direct access to state resources and low levels of loyalty toward the government leader. Also, the provisional nature of many power-sharing institutions increases rent- seeking incentives: facing a limited time horizon in oﬃce, rent-seeking elites within the power-sharing coalition are likely to capture as many rents as possible before they have to leave oﬃce. Thus, post-conﬂict countries with power-sharing institutions should exhibit higher aggregated levels of rent-seeking measured as the level of corruption in a country. In a statistical analysis of all post-conﬂict situations during 1996–2010, we ﬁnd that power-sharing cabinets substantively increase corruption in post-conﬂict countries and that this eﬀect is stronger in the presence of natural resource rents. These ﬁndings add quantitative evidence to the debate about drivers of post-conﬂict corruption. Moreover, they highlight a trade-oﬀ between short-term stability and long-term negative eﬀects of corruption for post-conﬂict political and economic development.


INTRODUCTION
Systemic corruption is an endemic problem in countries emerging from violent conflict (Cheng & Zaum, 2013;Lindberg & Orjuela, 2014). Higher corruption is negatively related to the quality of post-conflict peace (Global Peace Index, 2015), it increases the risk of renewed ethnic conflict (Neudorfer & Theuerkauf, 2014) and is generally assumed to have negative effects on long-term economic development (Aidt, 2003). The weakly institutionalized environments of many contemporary post-conflict states are an ideal breeding ground for practices that allow the use of public office for private gain-the common definition of corruption that we follow (Svensson, 2005). Given this substantive significance of corruption in post-conflict situations, we observe a surprisingly wide variation in corruption practices across postconflict states. Although all post-conflict countries score, on average, lower on corruption indices than countries without a history of violent conflict, there are stark differences in the level of corruption within the group of post-conflict countries. States such as Rwanda, Peru, or Croatia are less affected by systematic bribery, patronage, and clientelism than, for instance, the Democratic Republic of the Congo or Liberia. What accounts for this variation?
We argue that one of the main drivers of systematic postconflict corruption is the presence of political power-sharing institutions in a country. We conceptualize government elites in any post-conflict situation as individual rent-seeking agents that need to ensure the support of their key constituencies to remain in power (Bueno de Mesquita, Smith, Siverson, & Morrow, 2003). We expect power-sharing institutions-parti cularly cabinet-level, executive power-sharing institutionsto systematically shape these rent-seeking motives. Powersharing cabinets create political coalitions dominated by small circles of government and rebel elites with direct access to state resources and low levels of loyalty toward the government leader (Bueno de Mesquita et al., 2003). Also, the provisional nature of many power-sharing institutions increases rentseeking incentives: facing a limited time horizon in office, rent-seeking elites within the power-sharing coalition are likely to capture as many rents as possible before they have to leave office (Levi, 1989). Based on this logic, we argue that postconflict countries with power-sharing institutions are on average associated with higher aggregate levels of rent-seeking conceptualized and measured as the overall level of corruption in a country. We also suspect that power-sharing institutions are more likely to increase corruption when there are higher levels of resources that are easy to capture, such as natural resources or foreign aid.
We test these predictions with a statistical time-series crosssectional analysis of post-conflict situations during 1996-2010. Our dependent variable measures the extent of corruption in a country using data from the Worldwide Governance Indicators (Kaufmann, Kraay, & Mastruzzi, 2010). We rely on the Power-Sharing Event Dataset (PSED) (Ottmann & Vü llers, 2015) to capture the practices of executive power-sharing in * We thank Richard Batley, Walt Kilroy, Melissa Nelson, Fiona Nunan, Jonathan Rose, Andrea Ruggeri, Christian von Soest, Alaa Tartir, Remco Zwetsloot, and members of the GIGA Research Programme ''Peace and Security" for helpful comments and discussions. We are also grateful for the comments we received from the editor of World Development and two anonymous reviewers. We acknowledge support from the German Research Foundation (OT 494/1-1, ME 1701/7-1). Authors are listed in alphabetical order. Equal authorship is implied. All replication materials are available at the Harvard Dataverse (http://dx.doi.org/10.7910/DVN/ CDTH6H). Final revision accepted: July 2, 2017. the aftermath of negotiated settlements. In contrast to the large majority of existing data collections on power-sharing, PSED moves beyond the coding of the mere promises of power-sharing found in peace agreements. Instead, PSED explicitly records whether and when power-sharing between the government and former rebels has taken place in a postconflict period and when it was abolished again. PSED therefore provides us with time-variant data on actual rebel access to state resources. Consistent with our predictions, we find empirical support for our hypothesis that executive powersharing increases the level of corruption in a post-conflict country. Our analysis also finds this rent-seeking behavior to be more pronounced in countries with higher levels of natural resource wealth. Our empirical results are robust to a range of alternative model specifications and treatment selection models that account for the endogenous selection of powersharing governments.
This article makes several notable contributions. First, we advance the literature on power-sharing after civil conflict. Most scholars and practitioners assume that power-sharing functions by mitigating commitment problems between government and rebels (Hartzell & Hoddie, 2007). Building on this theoretical assumption, they explore how power-sharing affects civil war recurrence. In contrast, we shed light on the mechanisms through which power-sharing works by exploring the relationship between executive power-sharing and corruption using new data on the implementation and duration of these arrangements (Ottmann & Vü llers, 2015). We identify a specific channel through which such commitment may be secured: the buying off of violent actors and the distribution of spoils from the state. We also add quantitative evidence to the largely case-study-based research on post-conflict corruption (Cheng & Zaum, 2013;Lindberg & Orjuela, 2014). By statistically isolating the effect of power-sharing on corruption, we identify a major driver of corruption in post-conflict settings beyond single cases.
Second, we make a more general contribution to the broader study of the institutional determinants of corruption in developing countries. In this line of research, scholars typically investigate the long-term effects of structural changes in key institutions-such as the democratization of political institutions or improvements to the rule of law-on corruption (Billger & Goel, 2009;Iwasaki & Suzuki, 2012;Jetter, Agudelo, & Hassan, 2015). Taking this research further, we focus on an important subset of countries in the developing world, post-conflict countries, and show that even shortlived transitional power-sharing governments in these countries can have a substantial effect on corruption levels.
These academic contributions enable us to inject new empirical evidence into a critical policy debate. Notwithstanding its mixed track record (Binningsbø, 2013), power-sharing remains a popular conflict-resolution method for policy-makers at the United Nations and foreign ministries around the world. Most recently, it has been discussed as a key element of a prospective negotiated settlement to the Syrian conflict (Groarke, 2016). Executive power-sharing was also central in the recently failed peace agreement in South Sudan (Kindersley & Rolandsen, 2016). But our findings highlight an inconvenient trade-off inherent to these arrangements. Power-sharing between government and rebels might be necessary to halt continued bloodshed. At the same time, power-sharing is also a main driver of corruption in post-conflict countries. While this corruption might be a necessary element of securing the peace, it threatens the long-term perspectives of post-conflict countries. Policy-makers should therefore focus not only on how to design successful power-sharing deals but also on how a country can transition away from power-sharing once the risk of civil war recurrence has substantially subsided.

CORRUPTION IN POST-CONFLICT COUNTRIES
Scholars and practitioners increasingly recognize corruption as one of the key challenges faced by countries emerging from violent conflict (Cheng & Zaum, 2013;Lindberg & Orjuela, 2014;Transparency International, 2014;Working Group on Corruption and Security, 2017). This recognition builds on the insight that post-conflict countries are, on average, more corrupt than countries not affected by conflict (see left panel in Figure 1). Typically, the effects of corruption in postconflict contexts are characterized as a trade-off between short-term stability and long-term negative effects on political and economic development (Cheng & Zaum, 2013;Lindberg & Orjuela, 2014). Cheng and Zaum (2013, p. 10) observe that ''[w]hen the proceeds of corruption flow through [patronage] networks during a period of political instability, then these funds can be used to cement loyalties and relationships which can, in turn, help to stabilize the emerging post-war order." These expectations are in line with the more general finding that elites in corrupt countries can use oil rents to ''buy peace" by co-opting regime opponents (Fjelde & Soysa, 2009). In the long-term, however, rampant corruption can deepen intergroup mistrust, particularly if one societal group is perceived to benefit more from corruption than any other group. Citing evidence from Sri Lanka, Orjuela (2014, p. 760) reports substantial differences in corruption perceptions across ethnic groups. This finding is echoed by quantitative evidence showing corruption to be a driver of ethnic civil war onset (Neudorfer & Theuerkauf, 2014). More generally, the negative long-term effects of corruption in post-conflict countries stem from the insight that corruption hampers economic growth (d'Agostino, Dunne, & Pieroni, 2016;Serritzlew, Sønderskov, & Svendsen, 2014) and democratic development (Fukuyama, 2014, chap. 5;Warren, 2004), by entrenching established elites and their patronage networks.
These effects are particularly interesting since we observe much variation in the degree of corruption across postconflict countries (see the right panel in Figure 1). One potential driver of corruption is the influence of natural resource abundance on corruption in certain post-conflict contexts. Billon (2014) argues that natural resources, such as oil, diamonds, timber, or coal can feed post-conflict corruption, but this relationship is mediated by the quality of existing institutions and socio-economic inequalities (see also Rustad, Lujala, & Le Billon, 2012). This insight is linked to quantitative evidence on the conditional effect of resource income on democracy and conflict in the context of high-quality and/or high-capacity institutions (Basedau & Lay, 2009;Dunning, 2008;Ross, 2015).
Another variable that features prominently in the studies on drivers of post-conflict corruption is international engagement, particularly the sudden inflow of foreign aid and its impact on corruption. Aid in post-conflict contexts, argues von Billerbeck (2013, p. 82), implies a ''sudden availability of rents, combined with [. . .] distorted economic conditions [that] makes post-conflict settings rife with opportunities for corruption." In 2008, for instance, aid to post-conflict Liberia peaked at almost 200% of the national gross-domestic product, dwarfing the local economy. But it is not only reconstruction and humanitarian aid inflows that increase opportunities for corruption. The deployment of large-scale peace operations, too, is often accompanied by the establish-