Civil wars are violent, fatalities produced by civil wars observed after the end of the Cold War accounted for approximately 80 percent of all war’s fatalities in the 1990s. Understanding why and when they occur is important for policymakers and academics alike. This has led to an active research agenda that emphasizes rebel greed versus real grievances and the capacity of the national government to police its own territories.

The implication for understanding civil war onset is simple. Governments with the capacity to police their territories are unlikely to experience the onset of a civil war. When governments are unstable, nascent, are mishandling the economy or have yet to penetrate remote or rough terrain, insurgency is likely. This is a state capacity argument for civil war onset advanced in Fearon and Laitin’s seminal study.

However, these arguments miss that state capacity is variable within states over time. Consider the modern history of Iraq. Since its independence in 1932, Iraqi governments have ruled three disparate territories during periods of political instability, stunted economic development, and severe sanctions. All three conditions promote insurgency and civil war, though Iraq had experienced only one limited-duration insurgency prior to the first Gulf War. Two more insurgencies followed after the Gulf War when Hussein’s government was in a weakened state. Only after the Iraq War has the country seen the level of sectarian violence and civil conflict suggested by a state capacity argument.

In an article appearing online now at Journal of Peace Research, Doug Gibler and I argue that the case of Iraq illustrates the effect of the external threat environment on state capacity and civil conflict onset. War with Iran and rivalries with neighboring states like Kuwait, Saudi Arabia, and Syria stunted development in Iraq but increased the extent to which Saddam’s Hussein government penetrated Iraqi society. The civil conflicts observed in Iraqi history occurred prior to Iraq’s wars and rivalries, the concentration of power in Baghdad, or after the foreign-imposed regime change of the Iraq War.

We argue that external threats to territory increase the capacity of the state to penetrate society and deter the onset of a militarized challenge to its authority from within its borders. To that end, we conceptualize the capacity of the state as two-fold. First, we consider state capacity as, in part, the “connectedness” of the average citizen with the state. Second, we consider state capacity as the ability to implement national government policies by force if necessary.

We argue external territorial threats increase both types of state capacity. External territorial threats increase the connectedness of the average citizen to the state over other political groupings (e.g. ethnic group). This builds on longstanding arguments that external conflict makes groups more cohesive and unified internally. When the average citizen self-identifies as a member of the state under conditions of territorial threat, popular opinion is unlikely to tolerate dissent of the government’s authority. Absent the opportunity to win “hearts and minds” to offset the government’s material advantage at the onset of a rebellion, potential insurgents are unlikely to challenge the government’s authority.

Further, external territorial threats increase the capacity of the government to repress potential regime dissidents. In his book, Gibler makes this point explicit in showing the average territorial dispute corresponded with a predicted increase of over 100,000 troops in states under territorial threat. Territorial threats from land-contiguous neighbors elicit the focus on standing armies to secure the contested territory that make national governments better equipped to repress challenges to its authority before those challenges meet the operational threshold of a civil war.

We employ a two-stage research design to evaluate our hypotheses. We start first by operationalizing our two measures of state capacity. We operationalize “state connectedness” as the natural log of government’s share of the gross domestic product per capita in a country-year. If external territorial threats increase the connection of the average citizen with the state, we should expect the citizen to be more willing to consent to resource extraction and increased government spending as a share of society’s overall resources. We operationalize a state’s capacity to repress using the natural log of military personnel in a country-year.

Table I shows the consistent effect of external territorial threat on state capacity. States experiencing an armed inter-state conflict over territory see an increase in military personnel and government’s consumption of resources in the following year. Only territorial conflicts of the two conflict types have this consistent effect on state capacity. Table III suggests that the predicted effect of territorial conflict is an increase of about 2,000 soldiers and a 1.5% increase in the government’s share of society’s resources.

Since no country experiencing an armed inter-state territorial conflict saw an intra-state war in the following year, we substitute a general indicator of territorial claims from land-contiguous neighbors as our measure of territorial threat in the civil conflict models in Table II. The results are consistent with our expectations and much of what we know of state capacity models of civil conflict but suggest novel findings.

First, we hypothesized that the increase in state capacity decreases the risk of civil war onset even if governments may be willing to tolerate small-scale internal fights. We find that states with external territorial threats are less likely to experience civil war onset using Correlates of War data and are less likely to experience an intra-state armed conflict onset at lower levels of intensity using UCDP conflict data. We also found that the increase in state capacity that follows a territorial claim from a neighbor persists to a 30-year period after the claim has been resolved. States that have recently resolved a territorial claim with a neighbor are unlikely to experience civil conflict onset. The supplemental appendix shows the results from Table II are robust to multiple model specifications.

Our findings imply that the conditions that promote insurgencies are more likely to result in challenges to the government’s authority in regions that have been peaceful for an extended period of time. When the government has not faced an external challenge to its territorial integrity for a long time, the average citizens feels less connected to the state, the ability to repress dissidents decreases, or both. Conditions that promote insurgencies against the government may be stronger in regions where peace has been the norm in the neighborhood.