Abstract: Authoritarian regimes adopt nominally democratic institutions despite the inherent uncertainty that these institutions entail. What effects do elections and parties have on authoritarian regime stability? Are autocracies with parties and competitive elections more stable than their less electorally competitive counterparts? What functions can elections serve in non-democracies? This paper provides a naïve Bayesian informational theory of electoral institutionalization under authoritarianism. Democratic emulation, exemplified in parties and elections, enhances the chance of authoritarian survival by supplying the ruling elites with (1) information about the sources of support and opposition to the regime among the citizenry and (2) information to monitor intra-party mobilization capacity, loyalty and performance. Autocrats update their beliefs about the probabilities of support and opposition after each election. Over time, this cumulative learning helps them maximize their survival chances. From this basic theory, three observable implications are derived and subsequently tested: (1) party-based authoritarian regimes are more stable than more closed authoritarian regimes, and (2) on average, the higher the degree of legislative and executive electoral competitiveness within a regime and across regimes, the higher the probability of regime survival would be, and (3) the stabilizing effect of electoral competitiveness grows over time. To test these predictions, I use three functional forms, respectively: (1) product-limit Kaplan-Meier with stringent log-rank specification, (2) a modified version of the Cox proportional hazard model, and (3) Aalen additive risk estimation. Despite the limitations of the models in drawing solid causal inference, the findings lend support to the stabilization by institutionalization hypothesis.