Michael K. Miller's formal model of incumbent-led democratic breakdown and erosion (Journal of Politics, 2021) analyzes when democracy can be self-enforcing: an incumbent may abuse power to consolidate it, an opposition may rally protest or cry wolf, and imperfectly informed, partisan citizens decide whether to protest. Its comparative statics — power-sharing (non-winner-take-all) institutions, high media accuracy, and low partisanship typically promote democratic survival, while citizen support for democracy is protective in some parameter regions and destabilizing in others — remain largely untested against data, and competing formal mechanisms yield partially opposing predictions. The paper closes by calling for further empirical and formal investigation of how breakdown and erosion operate and by asking whether established institutional constraints, in particular those of the United States, will hold under strain. The agenda remains open: partial empirical tests of adjacent hypotheses exist, but no study has directly and systematically evaluated the model's institutional and media predictions, and the US question is unresolved.

Background

Since the end of the Cold War, the typical way democracies die has changed. Nancy Bermeo (2016) documented a shift away from open-ended military coups and blatant election-day fraud toward executive aggrandizement — elected leaders gradually dismantling checks on their own power while keeping a democratic facade — and strategic electoral manipulation. The gradual, incumbent-led variant is now usually called democratic erosion, as distinct from democratic breakdown, the outright end of competitive rule. Anna Lührmann and Staffan Lindberg (2019) used expert-coded democracy indices to argue that a third wave of autocratization driven largely by such incumbent-led erosion was underway.

Michael K. Miller (2021) built a formal (game-theoretic) model asking whether democracy can be self-enforcing — stable because elites and citizens each find it in their interest to defend it, rather than because of any external enforcement. An incumbent may violate democratic rules to consolidate power (a self-coup, i.e., an incumbent neutralizing constitutional constraints to expand its own power); an opposition can call citizens to protest, but can also cry wolf by calling protest when no violation occurred; and citizens do not observe violations directly, instead receiving private signals that equal the truth with probability qq (the model's measure of information or media accuracy), with partisan motivation distorting their response. Democracy survives in equilibrium only when the incumbent's current and expected future shares of office benefits, β\beta and β′\beta', are moderate: winner-take-all stakes — institutions in which one electoral winner controls nearly all power — tempt the loser to subvert the system, whereas power-sharing institutions (coalition cabinets, federalism, independent courts, and similar devices that divide power among actors) reduce the temptation. The model yields comparative statics, i.e., predictions about how parameter changes shift outcomes: non-winner-take-all institutions, higher media accuracy, lower partisan polarization, and stronger citizen support for democracy typically make democracy more likely to survive — but, strikingly, when media accuracy is sufficiently low, stronger citizen democratic values can instead make protest against a non-abusing incumbent more likely and can thereby destabilize democracy. Extensions let incumbents choose erosion (a partial, less detective power grab) rather than full breakdown, invest in degrading the information environment (fake news), and strategically inflame partisanship.

The model speaks to an empirical literature. Benjamin A.T. Graham, Miller, and Kaare W. Strøm (2017) had found, in a global dataset covering 1975–2015, that constraining powersharing institutions (such as independent judiciaries and civilian control of the military) consistently promote democratic survival, while inclusive and dispersive forms do so only conditionally. Milan W. Svolik (2019) argued that polarization leads voters to trade democratic principles for partisan gain; survey work by John Carey and colleagues (2022) shows many US partisans tolerate co-partisan candidates who undermine democracy; and Svolik (2023) estimated how much voters in fact punish subversion, through vote switching, backlash, and disengagement. What no study has done is evaluate Miller's model and its specific predictions directly and systematically — leaving both the empirical status of the strategic account and its application to established democracies such as the United States open.

Problem Statement

Following Miller (2021), the open question comprises three linked fronts of a single inquiry into how modern democratic breakdown and erosion operate. (1) Empirically: do the strategic model's predictions hold in data? Specifically, do non-winner-take-all (power-sharing) institutions, higher media accuracy, and lower partisan polarization promote democratic survival; is citizen support for democracy protective in some parameter regions yet destabilizing in others (as when low media accuracy makes stronger democratic values increase protest against non-abusing incumbents); and do incumbents in fact pursue gradual erosion rather than outright breakdown, degrade information environments, and strategically inflame partisanship, as the model's extensions imply? (2) Formally: how do these conclusions change as the strategic environment is enriched — oversight institutions such as courts and media, endogenous polarization, repeated interaction — given that competing formal mechanisms (citizen reference points, contested oversight, protest salience) yield partially opposing predictions? (3) As the model's own application: do established institutional structures — in particular those of the United States, with their combination of winner-take-all features, a polarized electorate, and declining trust in media — leave executive constraints binding in the model's terms, or does that configuration predict increasing strain and democratic backsliding? Miller's conclusion frames the agenda as an ongoing one — scholars should "continue to investigate (both empirically and formally) how modern democratic breakdown and erosion operate" — and he himself situates the US question in the model's terms (winner-take-all institutions, accelerating polarization, declining trust in news).

References

  1. Michael K. Miller (2021). A Republic, If You Can Keep It: Breakdown and Erosion in Modern Democracies. The Journal of Politics 83(1):198–213. DOI: 10.1086/709146. https://doi.org/10.1086/709146
  2. Nancy Bermeo (2016). On Democratic Backsliding. Journal of Democracy 27(1):5–19. DOI: 10.1353/jod.2016.0012. https://doi.org/10.1353/jod.2016.0012
  3. Anna Lührmann and Staffan I. Lindberg (2019). A third wave of autocratization is here: what is new about it? Democratization 26(7):1095–1113. DOI: 10.1080/13510347.2019.1582029. https://doi.org/10.1080/13510347.2019.1582029
  4. Benjamin A.T. Graham, Michael K. Miller, and Kaare W. Strøm (2017). Safeguarding Democracy: Powersharing and Democratic Survival. American Political Science Review 111(4):686–704. DOI: 10.1017/S0003055417000326. https://doi.org/10.1017/S0003055417000326
  5. Milan W. Svolik (2019). Polarization versus Democracy. Journal of Democracy 30(3):20–32. DOI: 10.1353/jod.2019.0039. https://doi.org/10.1353/jod.2019.0039
  6. John Carey, Katherine Clayton, Gretchen Helmke, Brendan Nyhan, Mitchell Sanders, and Susan Stokes (2022). Who will defend democracy? Evaluating tradeoffs in candidate support among partisan donors and voters. Journal of Elections, Public Opinion and Parties 32(1):230–245. DOI: 10.1080/17457289.2020.1790577. https://doi.org/10.1080/17457289.2020.1790577
  7. Edoardo Grillo and Carlo Prato (2023). Reference Points and Democratic Backsliding. American Journal of Political Science 67(1):71–88 (online 2021). DOI: 10.1111/ajps.12672. https://doi.org/10.1111/ajps.12672
  8. Milan W. Svolik (2023). Voting Against Autocracy. World Politics 75(4):647–691. DOI: 10.1353/wp.2023.a908772. https://doi.org/10.1353/wp.2023.a908772
  9. Marc S. Jacob (2025). Citizen support for democracy, anti-pluralist parties in power and democratic backsliding. European Journal of Political Research 64(1):348–373. DOI: 10.1111/1475-6765.12703. https://doi.org/10.1111/1475-6765.12703
  10. Andrew T. Little and Anne Meng (2024). Measuring Democratic Backsliding. PS: Political Science & Politics 57(2):149–161. DOI: 10.1017/S104909652300063X. https://doi.org/10.1017/S104909652300063X
  11. Robert Gulotty and Zhaotian Luo (2026). Backsliding under contested oversight. Political Science Research and Methods (FirstView):1–21. DOI: 10.1017/psrm.2026.10109. https://doi.org/10.1017/psrm.2026.10109
  12. Abel Escribà-Folch and Joan C. Timoneda (2025). Can Social Media Help Incumbents Subvert Democracy? Politics & Society 54(1):64–92. DOI: 10.1177/00323292251362676. https://doi.org/10.1177/00323292251362676
  13. Sebastian Schweighofer-Kodritsch, Steffen Huck, and Macartan Humphreys (2025). Political salience, endogenous bandwagoning, and regime resilience. Games and Economic Behavior 154:79–96. DOI: 10.1016/j.geb.2025.08.011. https://doi.org/10.1016/j.geb.2025.08.011
  14. Gabriele Gratton and Jacob Edenhofer (2025). The rise and fall of technocratic democracies: unstable majorities and delegation to technocrats. Journal of European Public Policy. DOI: 10.1080/13501763.2025.2576160. https://doi.org/10.1080/13501763.2025.2576160
  15. Anna M. Meyerrose (2025). Eroding Democracy from the Outside In. Oxford University Press. DOI: 10.1093/9780197816479.001.0001. https://doi.org/10.1093/9780197816479.001.0001