Arab political regimes are both unusually undemocratic and unusually stable. A series of nested statistical models are reported to parse competing explanations. The democratic deficit is comprehensible in terms of lack of modernization, British colonial history, neighborhood effects, reliance on taxes for government finance, and the Arab population share. Interpretation of the last variable is problematic: It could point to some antidemocratic aspect of Arab culture (though this appears not to be supported by survey evidence), or it could be a proxy for some unobservable such as investment in institutions of internal repression that may not be culturally determined and instead reflect elite preferences. Hypotheses that did not receive robust support include the presence of oil rents, the status of women, conflict with Israel or other neighbors, or Islam. The odds on liberalizing transitions occurring are low but rising. In this respect the distinction between the interpretation of the Arab ethnic share as an intrinsic cultural marker and as a proxy for some unobservable is important—if the former is correct, then one would expect the likelihood of regime change to rise only gradually over time, whereas if it is the latter, the probabilities may exhibit much greater temporal variability.
