We create a rotating bandit model of executive turnover in politics with au
tocratic presidents, large and centralized governments, and limited reelect
ion. The model is an extension of McGuire and O1son's 1996 work. We apply o
ur model by studying the relationship between electoral cycles and economic
growth and inflation uncertainty in Mexico, a country with a highly centra
lized and powerful government, no reelection, and until recently, little po
litical competition. We find a significant postelection economic collapse b
ut no preelection boom, which is contrary to the predictions of the traditi
onal political business cycle model. We also find evidence that elections c
reate, rather than resolve, inflation uncertainty, which contradicts the pr
edictions of the rational partisan model. While our rotating bandit model i
s largely consistent with the results we find, more work is needed on the r
eal effects of politics in the developing world.