Files
Abstract
When democracies decline, which institutions are most likely to be targeted first by would-be autocrats? I argue there is a standard sequence by which democratic institutions are undermined, at least in presidential democracies. Would-be autocrats in presidential systems with term limits (legal maximums on time in office) must circumvent them to succeed at their goal of staying in power; term limits therefore may incentivize would-be autocrats to erode some democratic institutions beforehand in order to optimize their probability of extending their legal tenure. Using a new and highly disaggregated global dataset of term limit evasions in presidential systems from 1990-2023 in conjunction with objective and subjective indicators of democratic decline across four institutions, I examine whether free assembly, legislative opposition, press freedom, and elections tend to deteriorate before would-be autocrats try to circumvent their term limits or after, and I discuss how the results can be interpreted as implying a 'relative' sequence of backsliding in presidential democracies. I find term limit evasion attempts are uniformly distributed temporally through president's legal tenures, rather than being concentrated early or late, implying it is particularly relevant to consider which preceding institutional assaults may serve as precursors to and products of evasion. I find that presidents who evade term limits have more frequently reduced the rights of protesters before their evasion attempt compared to after, and most evaders have partisan control of their legislatures. I also discuss institutions where there are low incentives for evaders to chronologize their attacks against other democratic institutions; press freedom tends to be as frequently undermined before as after evasion. Lastly, I expected to find elections deteriorate following term limit contravention, but there was no indication of post-evasion declines in electoral integrity. I discuss theoretical takeaways and limitations of identifying relative, rather than absolute, sequences of backsliding.