And fifth and finally, we trace a sharp increase in international conflict, visible in the surge in interstate wars in recent years, but especially pronounced when we widen the lens to internationalized conflict as well.

Fourth, more tentatively and with a delay that can be expected, we document a rise in civil conflict along ethnic lines.

Third, combining this data with information on transnational ethnic kin, we find that the number of irredentist configurations - state dyads with excluded kin abroad - have increased similarly for the first time in decades.

Mirroring the nationalist trend, we find that after decades of decline, the share of ethnic groups excluded from executive power has risen since 2010, with 65 previously included groups losing political access.

We document each of these steps using systematic data. First, using information from V-Dem, we show a steep rise in the share of governments relying on nationalist legitimation strategies, constituting a sharp reversal from trends of the initial post-Cold War period.

Causal sensitivity analyses indicate that our results are robust even to severe omitted variable bias. Difference-in-differences analyses that rely exclusively on over-time variation within groups yield consistent effects that tentatively support a causal interpretation.

Using this data, we find that groups subject to dominant nationalist demands face a 0.6-0.7% higher annual risk of civil war onset. This is substantial relative to the sample mean of 0.7% and comparable in magnitude to established explanations of civil war.

For these 90 countries, our data show that while the number of dominant nationalist movements has grown only slowly, their governmental influence has increased markedly since the 2010s, in line with observations of a nationalist resurgence around the globe.

We argue that dominant nationalism raises the risk of civil war through three mechanisms. First, it pushes governing elites toward the political exclusion of targeted outgroups, even where strategic incentives would favor accommodation.

An alternative analysis using continuous measures echoes this pattern: autonomy reduces mobilization when groups are demographically dominant in their region(s) and concentrated in a single region; however, this effect is reduced or reversed as dominance falls or fragmentation rises.

My main results (FECT): unified ethnic autonomy reduces a group's annual probability of separatist mobilization by ~10%. This effect grows over time and is robust across alternative estimators, thresholds, and measurement choices.

To test these arguments, I introduce new time-variant spatial data on internal administrative boundaries for 180 countries from 1946 to 2018, based on more than 8000 geo-coded regional polygons, combined with constitutional coding of which regions classify as autonomous.

I argue that the answer hinges on how closely regional boundaries align with ethnic settlement areas, depending on each region's ethnic make-up and each group's fragmentation across regions. To capture this, I distinguish three ideal-typical configurations of regional autonomy.