We empirically investigate the existence of spatial autocorrelation between military dictatorships in Sub-Saharan Africa from 1977 through 2007. We apply a Bayesian SAR probit regression, extended to a pooled model. We find a robust and positive spatial autocorrelation coefficient, which shows a spatial concentration of military autocracies. In particular, in the aftermath of Cold War military regimes cluster in the central region. Among covariates, interestingly, foreign aid shows a positive association with military regimes during the Cold War while it turns to exhibit a negative association after 1989. With regard to other economic covariates, we find that: a) there is a negative association between GDP per capita and the existence of a military autocracy; b) a larger manufacturing sector is associated with a smaller probability of a military rule; c) a larger mining sector is associated with a higher likelihood of military rules; d) trade openness reduces the likelihood of militarization.
military dictatorship Sub-Saharan Africa Bayesian SAR probit model spatial autocorrelation diffusion concentration