No abstract

Territorial autonomies are increasingly implemented as tools for regulating ethnic conflicts. Since there are successful as well as failed cases, the ongoing debate about a general conflict-solving potential is not a very fruitful one. The article turns towards the analysis of various factors affecting the success of conflict regulating autonomies. It is argued that successful conflict regulation highly depends on the scope of self-rule transferred to a rebelling group. It offers two approaches to measure the scope and success of autonomies and introduces the Autonomy Scope Index (asi). The relevance of the degree of autonomy is tested on the basis of a global data set on all conflict regulating autonomies since 1945 by using Qualitative Comparative Analysis (qca). The study shows that a substantial degree of self-rule is indeed a crucial condition.

Territorial autonomies (TA) are increasingly implemented as tools forregulating ethnic conflicts. Since there are successful as well as failed cases,the ongoing debate about a general conflict-solving potential is not a veryfruitful one. The article turns towards the analysis of various factors affectingthe success and failure of conflict regulating autonomies. It is argued thatautonomy consolidation requires specific conditions which support theprocess of mutual recognition between ethnic groups. Two causal models aredeveloped and tested on a global data set on all conflict regulatingautonomies using a two-step Qualitative Comparative Analysis (QCA). It isfound that a combination of low horizontal inequalities, a high scope ofautonomy, inclusive institutions and international support lead to autonomysuccess whereas persistent inequalities, a low scope of transferredcompetencies and exclusive institutions result in a return to violent strategies.One important implication of these findings is that favorable conditions forpost-conflict institutional engineering can, at least to some extent, be created.

Online Appendices* p < 0.10; ** p < 0.05; *** p < 0.01 Results from ordered probit regression. Unstandardized coefficients with robust standard errors clustered by country-year. Weighted by ESS pweight and pspweight. * p < 0.10; ** p < 0.05; *** p < 0.01 Results from OLS regression. Unstandardized coefficients with robust standard errors clustered by country-year. Weighted by ESS pweight and pspweight. * p < 0.10; ** p < 0.05; *** p < 0.01 Results from OLS regression. Unstandardized coefficients with robust standard errors clustered by country-year. Weighted by ESS pweight and pspweight. * p < 0.10; ** p < 0.05; *** p < 0.01 Results from OLS regression. Unstandardized coefficients with robust standard errors clustered by country-year. Weighted by ESS pweight and pspweight. * p < 0.10; ** p < 0.05; *** p < 0.01 Results from ordered probit regression. Unstandardized coefficients with robust standard errors clustered by country-year. Weighted by ESS pweight and pspweight. * p < 0.10; ** p < 0.05; *** p < 0.01 Results from multinomial logistc regression. Unstandardized coefficients with robust standard errors clustered by country-year. Weighted by ESS pweight and pspweight. "Allow some" (3) is the base outcome. * p < 0.10; ** p < 0.05; *** p < 0.01 Results from multinomial logistic regression. Unstandardized coefficients with robust standard errors clustered by country-year. Weighted by ESS pweight and pspweight. "Allow some" (3) is the base outcome.