Do wars strengthen or weaken democratic institutions? Classic accounts point in opposite directions. For example, Tilly (1992) argues that, in 17th and 18th-century England, repeated interstate wars forced the state to raise taxes and bargain with Parliament, strengthening parliamentary authority and tightening constraints on the executive. In contrast, Tocqueville (2011) argues that, in Revolutionary France, the combination of an intrastate conflict (the Revolution) with a series of interstate wars led to a large expansion in executive power. More recently, Becker et al. (2025) provide evidence on the consequences of conflict on the development of democratic institutions in Europe, and Vargas et al. (2019) show that political violence in Colombia shapes local fiscal institutions.

Which of these paths prevails is ultimately an empirical question. In a new paper (Benmelech and Monteiro 2026), we provide the first systematic global evidence on when, where, and why conflict erodes democratic institutions. We do so by leveraging the dataset constructed for an earlier paper of ours Benmelech and Monteiro (2025), which includes 115 conflicts and 145 countries over the past 75 years, including both interstate wars (state versus state) and intrastate wars (state versus non-state).

A global view: Democracy around war

Our analysis combines comprehensive conflict data from the UCDP/PRIO Armed Conflict Dataset
with high-resolution institutional measures from the Varieties of Democracy (V-Dem) project. We track how democratic institutions evolve before and after the onset of conflict, comparing countries that enter war (treated countries) to otherwise similar countries that remain conflict-free (control countries). Our primary measure is the V-Dem Democracy Index, which captures multiple dimensions of the quality of democratic institutions.

Two facts stand out immediately. First, countries do not enter war because they are becoming less democratic. In the years before the onset of conflict, democratic institutions in treated countries are stable – or even improving slightly – relative to control countries. Second, autocracies are no more likely to engage in conflict than democracies in the same region.

Conflict causes a large and persistent drop in the quality of democratic institutions, as measured by the aggregate democracy index. As shown in Figure 1, the onset of conflict is associated with a 3% drop in democracy. More strikingly, the decline continues for nearly a decade, even though the median conflict lasts only three years. Ten years after the onset of conflict, democracy in treated countries is about 13% lower than in control countries. This is a large decline – the estimated ten-year change lies in the 14th percentile of the global distribution of decade-long changes, meaning that 86% of all observed shifts in democracy worldwide are less negative than our estimated effect.

Figure 1 Effect of conflict on the democracy index

The decline in democracy unfolds through a coherent set of political channels. Conflict leads to large and persistent increases in media censorship and judicial purges, directly weakening constraints on executive power. Core civil liberties also erode: freedom of association declines, and governments come to rely more heavily on the military as a base of political support. These changes are accompanied by greater political instability. Treated countries are more likely to experience extra-legal leadership turnover – such as coups, forced resignations, or assassinations – and to suspend constitutional rules.

Taken together, these patterns point to a common mechanism. The institutional changes triggered by conflict are political rather than functional: they weaken opposition, relax checks on the executive, and expand coercive authority, but do not reflect the operational requirements of fighting a war. Instead, conflict creates opportunities for incumbents to reshape institutions in ways that would be difficult to justify in peacetime – and whose effects persist long after the fighting ends.

Not all wars erode democracy

Democratic backsliding is not universal. It is concentrated in very specific political and institutional environments.

First, democratic erosion occurs almost entirely in first-time conflicts. Countries that have fought wars before experience little additional institutional damage. In contrast, first-time conflicts lead to large and persistent declines in the quality of democratic institutions. Once democratic checks have been weakened, there is simply less left to erode.

Second, backsliding is driven by intrastate conflicts, not interstate wars. Civil wars and internal conflicts – where governments face domestic challengers – lead to sharp institutional deterioration. External wars do not. Moreover, democratic decline following the onset of intrastate conflict is concentrated in highly fractionalised societies, where internal divisions magnify the political returns to repression.

Finally – and most strikingly – only countries that win wars experience democratic backsliding. Losers do not. As we show in Figure 2, a decade after the onset of conflict, the quality of democratic institutions in winning countries is nearly 40% lower than in the control group.

Figure 2 Winners versus losers

War does not reward autocracy

A natural interpretation is that democratic erosion reflects the functional demands of war: perhaps centralised authority is simply more effective in combat. This view is consistent with our findings in Figure 2 – maybe countries win the war because they have become more autocratic.

The data reject this explanation. We show that countries that become more autocratic during conflict are no more likely to win wars. If anything, institutional instability reduces the probability of victory. Therefore, we find no evidence that a declining democracy is a functional requirement of war.

War as a political opportunity

The evidence instead supports a political interpretation. Conflict reshapes domestic politics by weakening opposition, expanding coercive capacity, and legitimising extraordinary measures. None of these phenomena is a functional requirement of war. Moreover, there is little evidence to support the hypothesis that these changes in the quality of institutions increase the likelihood of victory. Victory further strengthens incumbents, allowing them to implement institutional changes that would be difficult – or impossible – in peacetime.

This helps reconcile competing views in the literature. War can build state capacity, but it can also undermine democracy. Whether it does so depends not on military necessity, but on political incentives.

Recent work (Acemoglu et al. 2025) has emphasised that democratic institutions are self-reinforcing when they function well. Our results show the darker counterpart: when war weakens those institutions, the damage can persist for decades.

Implications

The lesson is not that war inevitably destroys democracy. Rather, democratic resilience depends on political constraints, not on the absence of conflict itself. First wars, internal wars, and victories are precisely the moments when those constraints are most vulnerable.

As armed conflict becomes more frequent worldwide, understanding how to prevent war from becoming a gateway to autocracy is not just a political concern – it is central to the long-run health of democratic and economic institutions.

References

Acemoglu, D, N Ajzenman, C G Akosy, M Fiszbein, and C Molina (2025), “(Successful) democracies breed their own support”, Review of Economic Studies 92(2): 621–55.

Becker, S O, A Ferrara, E Melander, and L Pascali (2025), “Wars, taxation, and representation”, VoxEU.org, 17 July.

Benmelech, E, and J Monteiro (2025), “The economic consequences of war”, NBER Working Paper No. 34123.

Benmelech, E, and J Monteiro (2026), “War and democratic backsliding”, NBER Working Paper No. 34734.

Tilly, C (1992), Coercion, capital, and European states, AD 990–1992, Cambridge University Press.

Tocqueville, A (2011), The old regime and the revolution, Cambridge University Press, originally published in 1856.

Vargas, J F, J Shapiro, A Steele, and R Ch (2019), “Death and taxes: Political violence shapes local fiscal institutions and state building”, VoxEU.org, 29 January.