Sam Seitz

Democratic peace theory, the idea that democracies tend to be less bellicose than non-democracies, has been called “the closest thing we have to an empirical law in the study of international relations” by political scientist Jack Levy. It has also played an instrumental role in shaping recent US foreign policy, contributing both to Clinton’s decision to expand NATO and Bush’s choice to confront the “Axis of Evil.” But does it stand up to scrutiny? The short answer is yes. Democracies are significantly more pacific – at least in the modern age – than their autocratic peers. But the case for democratic peace theory is far from ironclad. Indeed, the theory faces three significant problems. The first relates to scholars’ inability to convincingly isolate the ways in which democratic regimes temper bellicosity in foreign policy. Democratic peace theory is strangely underdeveloped in this regard, and there is still an ongoing debate over the specific causal mechanisms linking democratic regime types to peaceful foreign policy. The second problem concerns measurements of democracy. These measurements are, unsurprisingly, awfully subjective, resulting in potential measurement errors. They are also difficult to benchmark across time. Finally, democratic peace theory has some major shortcomings when applied to specific cases. So, while democracies do tend to be more pacifically minded, under certain conditions they can be just as aggressive (and sometimes even more aggressive) as states led by autocratic regimes.

How Does Democracy Constrain?

What is it about democratic regimes that results in more pacific foreign policies? The most obvious response would be to highlight the role of regular elections. Stupid wars will lead to electoral punishment, as citizens have no desire to send their children off to war and suffer tax hikes to boot. There is some evidence supporting this explanation: democracies fight fewer wars, and they tend to win the wars they enter. This suggests that democratic leaders are more careful when deciding whether to embark on a path to conflict. Indeed, there is an additional advantage that democracies have, and that is that leaders of democracies can be peacefully swapped out via regular elections. Autocratic leaders, by contrast, face death or imprisonment upon their ousting, so they are incentivized to gamble for resurrection and escalate ill-conceived wars in the off-chance that they catch a lucky break and manage to parlay victory into greater personal security. Of course, the desire to oversee a successful foreign policy exists among democratic leaders as well, and even democratic leaders tend to get locked into certain paths regardless of their desirability. Still, democratic leaders’ motives are less existential and thus more easily tempered by angry voters and dissatisfied elites than those of desperate dictators trying to avoid a bullet to the head.

While these explanations may seem convincing, they face important challenges. Perhaps most significant is that institutional constraints are not unique to democracies. As selectorate theory argues, even autocrats face constraints from other elites. This is true to greater and lesser degrees, and true tyrants in the vein of Stalin do probably face relatively few limits to their freedom of action, but most autocrats must still appease the oligarchs and military leaders of their country if they seek to retain power. Khrushchev and Gorbachev offer cautionary tales to those who would believe that dictators are immune to elite politics, as both were the victims of coups that resulted from elite dissatisfaction at their handling of the Cuban Missile Crisis and economic reform, respectively. In other words, the constraints facing democratic leaders frequently fetter autocrats as well, and at least some evidence suggests that institutionalized autocracies are more peaceful than even democracies. If this is the case, democracy might be overrated, with its only potential advantage being that it might better prevent institutional backsliding.

The institutional explanation also faces a problem in that it assumes a level of voter engagement that may be unrealistic. We have all seen the clips of uninformed citizens failing to identify the United States on a map and spouting insane gibberish about all manner of things, so it is difficult to believe that these same people can be trusted to reliably reign in committed war hawks intent on leading the nation to war. Of course, eventually the damage to lives and economic growth becomes so extreme that even the most politically disengaged person is forced to have an opinion, but this can take years or simply never occur in the case of a small, low-casualty conflict sustained by government borrowing instead of taxes. Indeed, clever demagogues can rile up the myopic masses with promises of glory and profit in a way that simply isn’t possible in an autocratic system regulated by highly informed elites. In the post-9/11 environment, for example, it was actually politically costly to oppose the War in Iraq (what, do you support the terrorists?) even though the invasion was poorly planned and, certainly in hindsight, a very bad decision.

All of this begs the question of whether elections actually do meaningfully serve to constrain and forces us to wonder whose voices really matter, and this opens up the institutional explanation to yet another set of attacks. After all, maybe it isn’t democratic institutions per se that matter but rather the people that they empower. If democratic institutions give voice to merchants and free traders, the democratic system might allow a powerful constituency for peace to develop and constrain military adventurism. But mercantilists might use their influence to lobby for aggressive expansion in pursuit of lucrative colonies and export markets, using tariffs to finance their aggressive preferences. Of course, this itself begs the deeper question of whether democracy is simply epiphenomenal. If democracy only emerges in economically dynamic, peaceful states, then democracy as a variable is spurious. It’s the economic interdependence and sociological underpinnings of market democracy (legal/contractual norms, for example) and not the salutary effects of democratic institutions that result in democracies being more peaceful.

Returning to the question of whose voices matter most, it’s also worth pondering if gender matters. Some recent work suggests that women tend to be less inclined to militarized solutions and thus more resistant to reckless wars. Thus, the expansion of voting rights or the implementation of quotas designed to boost the number of women in positions of executive leadership might be more than democracy itself. If this is the case, it isn’t democratic checks that matter so much as the inclusion of female perspectives. Perhaps it is for this reason that one observes the democratic peace becoming much clearer post-WWI, as this was a period in which women became enfranchised in much of North America and Western Europe. Of course, it is important not to essentialize gender preferences, and it is also worth noting that several studies (see here, here, and here) have shown that female leaders tend to be more aggressive (or at least no less aggressive) than their male counterparts (after all, Golda Meir and Margaret Thatcher were hardly dovish). Nevertheless, these observations suggest that the kinds of voters in a democracy, rather than democracy itself, are key to explaining democratic states’ proclivity for peace.

But perhaps institutions do matter after all, and it is simply the case that the particular institutional explanations adumbrated above are incorrect. For example, maybe it isn’t elections per se that matter but the signaling potential they offer. Put differently, the transparency and open deliberative process of democratic regimes plausibly make their commitments more credible and their threats more serious. Adversaries can gauge the level of public support and the degree to which there is political buy-in, which might enhance deterrence and reduce the possibility of miscalculations leading to war. The ability to generate organic, domestic support also makes democracies potentially potent foes, as democratic regimes may be able to more fully mobilize for victory. Finally, democracy’s open, deliberative approach to governance might reduce the possibility of treaty violations or military provocations by making it more difficult for a regime to carry out a bluff without being exposed and criticized by domestic opposition groups. More assured in the credibility of a democracy’s commitments, adversaries might choose diplomacy over war. In short, perhaps it is not elections that matter so much as the open and free debating of ideas.

Still, this explanation seems incomplete. Saddam, for example, did not believe US threats in the leadup to the First Gulf War. This was partly because he felt democracies are weak and casualty averse, so he could simply outlast the US in a war of attrition. But his beliefs also resulted from the relatively lackluster public support for the war. Of course, everyone was for the war once it proved a success, but there was deep skepticism in the months before combat began, which of course proves that voters are just as prone to error as strongmen. And while democratic systems do allow for more credible signaling, it is worth acknowledging that just in the past few years Donald Trump has exploited his power to unilaterally withdraw the US from Senate-ratified treaties and agreements. Finally, these types of signaling arguments fail to explain why democratic peace theory seems to be dyadic. Put differently, democracies tend not to fight other democracies, but they actually do fight non-democracies at a fairly significant rate. If all that mattered was credible signaling and domestic mobilization, one would expect democratic peace theory to apply more broadly. So why doesn’t it?

One explanation emphasizes normative and ideational factors, suggesting that democratic states feel a sort of kinship with each other that militates against violent conflicts. For all the frustrations Americans have with their allies (and vice versa), anyone suggesting war with one of America’s democratic partners would be ridiculed and condemned by the American public. This has been tested experimentally, and the results are quite powerful. Ceteris paribus, American and British voters are extremely averse to initiating war against a democratic state, even if it is pursuing fairly destabilizing or provocative actions. The results are not quite conclusive, as the study neglects to poll people from non-democracies (in other words, it fails to prove that the affinity for democratic states is unique to citizens of other democracies), but it still offers powerful evidence suggesting that democratic voters do not want to fight other democracies. And yet there are still issues. What counts as a democracy? One could make a somewhat credible (though in my view fallacious) argument that the US is not really a democracy due to the absurdity of the electoral college, the disenfranchisement of felons, credible accusations of voter suppression in the south and gerrymandering pretty much everywhere. Indeed, one study shows that the citizens of Entente powers increasingly downgraded the democracy rating of Imperial Germany, which had a weak elected parliament, in the leadup to the Great War. If voters can shift their perceptions of other countries’ level of democracy so quickly, just how durable are these alleged bonds of shared identity?

To be clear, there is very obviously something about democracies that induce moderation and more peaceful behavior, at least toward other democracies. The problem is that we still don’t have a good idea of what it is about democracies that leads to this behavior. This makes it difficult to create policy, because different explanations suggested different priorities. If executive constraints are key, democracy isn’t even necessary – institutionalized autocracies would be just fine. If it’s important to have a strong merchant class, ingrained norms of property rights, and a legalistic culture, democracy is perhaps neither necessary nor sufficient. If women are the key, democracy might be necessary but insufficient. The point is not to argue that democratic peace theory is wrong, because it isn’t. But it is incomplete.

Measuring Democracy

The previous section addresses just the first problem with democratic peace theory, yet two further problems remain. I’ve already hinted at the measurement problems above, but it’s worth providing a bit more detail regarding just how biased and subjective some of the democracy scores are. For example, Polity (one of the main measures of democracy whose scores ranges from 10 to -10) moved the US and UK from a perfect ten to eight after 2016. That seems suspiciously like partisanship to me – the coders were anti-Brexit and anti-Trump and therefore conflated losing an election with losing democracy. The data become more suspicious when one realizes that the Nixon administration’s chicanery engendered a drop in the US Polity score but FDR’s court packing did not. That is awfully suspicious! Interestingly, Poland and Hungary are both coded as perfect democracies despite the flawed and rigged electoral and media systems that exist in those countries. To be blunt, anyone who thinks that Hungary is more democratic than the United Kingdom or United States has no business being in this profession.

Of course, the intertemporal comparisons also yield bizarre results. The US today now holds a lower score than the US of 1845 (when slavery was legal and women couldn’t vote) or the US of the 1920s (when Jim Crow still existed and the political machines manipulated American democracy). This is clearly absurd. Of course, one might argue that this approach to coding is superior because it benchmarks scores to the average democratic levels of the time, but that simply cannot be true. Hungary is not a full democracy by the democratic norms of this period, for example. And even if it were true, this kind of benchmarking would pose problems for quantitative studies that look at the role of democracy in constraining countries’ propensity to engage in warfare. After all, institutional constraints are not exactly relative. You either have a broadly enfranchised population that can hold politicians accountable and send credible signals or you don’t. Obviously 19th century democracies had slightly stronger institutional checks than their autocratic and monarchical peers, but they were not nearly as robust as those that exist in a fully democratic modern state. So, when one does quantitative analysis and assigns the same score to 1840s America and 2010s America, the results will be biased. Garbage in, garbage out.

This suggests a broader problem, though, and that is that many of these coders are biased by their place of origin. This is not disinterested coding but politically motivated assessment that is extremely hard to verify. The complexity is exacerbated by the manner in which these scores are calculated, as the overall score is an amalgam of several different constituent scores that are weighted strangely and are thus hard to objectively evaluate. This also means that disaffected academics upset about an election or political decision can lower a country’s score and then concoct some contrived post hoc justification.

One final problem is that the coders don’t seem to know what they want these scores to represent. It seems to me that some of the decisions regarding the US are arguably justifiable if one is just conducting intra-country assessments: The DOJ under Barr is clearly corrupt, and Trump has generally faced few constraints (especially early on in his presidency). It is therefore not crazy to argue that American democracy today is under more strain than it was during 2010. But most people use these scores to do inter-country comparisons, and so downgrading the US in practice means that these coders are saying that the US has declined relative to other countries, not relative to itself. And this is a problem because it leads to absurdities like Iraq and the US being coded as possessing the same level of polarization. One of these countries has been suffering civil wars and terrorist insurgencies for a decade and the half and the other has had some brief government shutdowns over budget battles. Perhaps I’m biased by my fondness for the United States, but those do not seem remotely equivalent to me.

These problems aren’t enormous because the level of bias is generally fairly moderate. And, of course, every kind of measurement of complex, somewhat normative phenomena will have a degree of subjectivity. Moreover, Polity is far from the only democracy index, so it is certainly possible to do robustness checks and mitigate the risk of measurement errors distorting results. But the broader problem of treating democracy as an integer value exists irrespective of the dataset used. There is no objective way to justify one country moving from 10 to 9 or 7 to 6. Consequently, many studies now use democracy dummies that employ an arbitrary cutoff value (usually 7 in my experience) to compare democracies versus non-democracies. Of course, this cutoff point is not objective either, but at least this approach prevents absurd regression outputs that suggest a “one-unit increase in democracy” results in a certain decrease in bellicosity. The upshot is that democracy is a concept that is hard to measure, so always treat quantitative studies on the subject with a degree of skepticism.

What the Theory Cannot Explain

The final problem with democratic peace theory is that there are several major exceptions that it struggles to explain. The biggest problem, at least in my view, is that new democracies act exactly opposite to the predictions of the theory. Far from reigning in their aggressive impulses, new democracies tend to find themselves in more wars than other states. This is not as damning as it may seem, because the research actually shows that it’s only during the transition to democracy that countries tend to experience violent conflict at much higher rates. Once democracy is established, young democracies are about as pacific as their mature democratic peers. This is not that surprising given work that shows that revolutionary regimes can be destabilizing and spark fear in their neighbors. Other work suggests that major shocks in the international system lead to corresponding waves of regime changes, and this unstable period might plausibly contribute to more inter- and intra-state violence. Still, the fact that specifically democratic transitions tend to me more violent is curious and suggests the dangers of demagoguery that exist in democratic regimes.

The theory also struggles to explain the behavior of the United States, a country that has engaged in near endless warfare throughout its history. The American Indians, European powers, and a range of smaller proxy and rogue states around the world have all experienced the sting of American military power at some point or another. In the roughly thirty years since the end of the Cold War, the US has been at war during about half of them. But this would seem strange, as the US has always been a democracy (though, as we’ve already discussed, a flawed one). Perhaps the answer to this paradoxical finding is that regime type is actually not that important after all. Yes, more democratic regimes tend to be more peaceful on the margins, but ultimately it’s the structure of the international system that matters. The US is the hegemonic power, and thus it seeks to dominate and control the system to ensure that its preferences dominate. If this were true, we’d expect the US to not fight other democratic states, as they were allies against the threatening Soviets and today serve to broadly support America’s international priorities. That they are democracies is purely coincidence, not causal. We’d also expect the US to be engaged in near constant warfare against marginal states that attempt to defy its rule and pursue strategies (like nuclear proliferation) that weaken American power. Of course, this is an extreme version of the argument, but it’s probably not totally incorrect. The balance of power matters, and the US—like every great power—seeks to retain its advantages, democracy or not.


The point of this post is not to downplay the power and importance of democratic peace theory. It is quite clear that democracy is strongly associated with more peaceful foreign policy, especially toward other democracies. And yet, the theory has always seemed underspecified to me. We know democracies fight each other less frequently than other states, but we simply do not have a conclusive explanation for why. And every proposed causal mechanism faces notable shortcomings. Additionally, much of the quantitative work relies on biased and imperfect indices. This is not disqualifying because the coding decisions are not so biased, but the potential issues are significant enough to raise skepticism over certain results and findings. Finally, there are some major challenges to the theory, including the foreign policy of the United States. Of course, US foreign policy adventurism does not itself undermine democratic peace theory, as most versions of the theory only suggest that democracies avoid fighting other democracies. But many of the mechanisms proposed as explanations of this phenomenon are not so explicitly dyadic, again raising doubt about our ability to explain the phenomenon of democratic peace. As with most academic theories, more work is required.