that the costs of aggression are too high for it to be worthwhile. Extended deterrence is a form of deterrence put into practice by a defender that wishes to prevent external aggression against another state – typically, a treaty ally. A defensive military alliance need not have to demonstrate that it would win a potential war for (extended) deterrence to work, but it must somehow convey that it has the political will, military capability, and resolve to inflict unacceptable pain on an adversary that it believes to be against the status quo (Jervis 1979). Deterrence is admittedly a deceptively simple concept. Just because an adversary has not acted aggressively does not necessarily mean that deterrence has worked. The adversary may never have had a desire to attack, making successful deterrence an illusion and thus a false positive (Lebow and Stein 1995). Divining adversarial intentions can be tricky, and with the stakes being sometimes existential, countries feel that they must err on the side of caution to undertake measures, like forming military alliances, to augment deterrence. And finally, deterrence does not imply having a militarily defensive strategy. Sometimes going on the offence is the best way to impose costs if deterrence fails.
Balancing Threat as a Classic Explanation of Alliance Formation
But what exactly is the danger that drives alliance formation? A classic explanation for why states establish defensive military alliances is that they are seeking to balance power. According to balance of power theory, states tend to band together so as to prevent a potential hegemon – that is, a state forecast to wield a preponderance of military and economic capabilities – from rising up and dominating the system (Levy 2004: 37). A military alliance can help reinforce the status quo and thus provide an invaluable tool for preserving it.
At first blush, this argument is intuitive and plausible. The emergence of Napoleonic France and Nazi Germany pushed states to unite in common cause and to aggregate their capabilities to defeat such bids for supremacy. On closer inspection, however, balance of power theory explains very little. Empirically, too many states across world history have been able to dominate regional systems by defeating rival coalitions (Wohlforth et al. 2007). Even in the foregoing cases, countries did not forge strong alliances (or, more accurately, military coalitions) against Napoleonic France and Nazi Germany until after those aspiring hegemons pursued conquest through war. Seven coalitions were attempted before Napoleonic France was defeated for good. Critically, the most powerful state in international politics in 2022 – the United States – has far more alliances than its potential rival China, which only has North Korea as a formal treaty ally. Balance of power theory, in its simplest incarnation, cannot explain the international environment in the early twenty-first century (Brooks and Wohlforth 2008). If anything, since the Cold War ended, there have been many more instances of bandwagoning (or joining the stronger side) with the United States than there have been attempts to balance against it.
Scholars have thus tried to refine balance of power theory to explain alliance formation better. Most notably, Stephen Walt (1987: 5) argues that states appraise others on the basis of perceived threat, which can be a composite of factors that include geography, perceived hostile intent, and offensive capabilities. In a twist on this argument, Paul Poast (2019) contends that compatibility in war plans and availability of outside options (buck-passing, unilateral action, or another alliance) drive alliance formation. Thomas Christensen and Jack Snyder postulate that if prevailing military technology favors the offence over the defense, then, all things being equal, states will be more likely to ally with one another against a common threat. The reasoning is simple. When defense is relatively easy, states might believe that others can do the fighting themselves against a perceived threat. Alliances do not happen – or at least will remain very weak – because of such “buck-passing” and “free riding.” However, if attacking is easier than defending, then the balance could be more easily overthrown. And so, “to uphold the balance and to have an effect on the outcome of the fighting, policymakers [believe] that they [have] to conclude binding alliances in advance and throw their full weight into the battle at the outset” (1990: 148). To be sure, according to Christensen and Snyder, such imperatives are most acute when more than two great powers among the states are present in a regional or international context – that is, under multipolarity. Otherwise, under bipolarity, whereby there are only two great powers of roughly equal standing, alliances should be much less consequential for actual war-fighting because no one ally or group thereof can tip the balance.
These refined arguments regarding the influence of the military balance have themselves come under criticism. The problem with Poast’s argument is that alliances often do not agree on actual war plans until sometime after they are established and after even more negotiation, if ever at all. How, then, can compatibility in war plans predict alliance formation but not actual alliance war plans? Moving on, some scholars take issue with whether we can neatly categorize weapons systems in an offence–defense binary – a point we will revisit later when we discuss theories of entrapment. In arguing against Snyder and Christensen, James Morrow asserts that states face a trade-off in choosing between establishing an alliance (i.e., external balancing) and developing one’s armaments (i.e., internal balancing). An alliance can enhance a state’s sense of security relatively quickly, but at the cost of sacrificing certain interests and flexibility to gain the cooperation of a prospective ally that may or may not come to its aid in an actual crisis. Building armaments gives the state a more reliable source of military power – its own – but at the political expense of allocating resources away from other uses. States manage this trade-off by selecting some mixture of arms and alliances, trying to optimize between them such that the costs of procuring either do not exceed their benefits. What determines these costs is usually domestic politics (Morrow 1993: 216). Acquiring armaments often hinges on how much the state is able to extract from society in order to meet its pressing security needs. Typically, great powers can mobilize enough manpower and military capabilities – hence their greatness – so they have a lesser need for alliances. For smaller states, there might exist domestic impediments to forming a military alliance. Though only small relative to the United States, Saudi Arabia is an illustrative example: anti-Americanism is sufficiently strong among members of its society that its leaders may be disinclined to negotiate an alliance treaty with the United States (Pollack 2003: 37). Whether such a trade-off exists between alliances and arms remains debatable: Morrow (1993: 214) acknowledges that allied states are not toothless, and an ally without arms would not be very desirable from a strictly security perspective (see also Horowitz et al. 2017). The proposition that states rationally allocate resources between arms and alliances would surprise those familiar with how dysfunctional military procurement efforts can be (Alic 2007: 107).
The notion that states face such a trade-off does highlight the importance of domestic politics for military alliances. It is perhaps no accident that the Cold War saw two competing superpowers – the Soviet Union and the United States – form rival military alliances on the basis of ideology, with communist countries grouped in one camp and anti-communist countries grouped in another. One study finds that, since 1945, states with similar regime types – whether they are both democratic or both autocratic – are more likely to ally with one another (Lai and Reiter 2000). Other statistical analyses have had trouble uncovering a correlation between alliance formation and regime type, with NATO and the Warsaw Pact being anomalies in the historical record (Simon and Gartzke 1996). NATO itself was mostly democratic but it did contain nondemocracies (Greece between 1967 and 1974 and Turkey at odd junctures) during the Cold War. Nevertheless, even if countries with different political systems form alliances, domestic politics, and, for that matter, ideology, can still matter. Some countries allied with the United States during the Cold War may not have been democratic, but their leaders still shared the same fears regarding communist expansion and subversion. Syngman Rhee was one such leader: although autocratic in method and disposition, he expressed anti-communist ideals long before becoming South Korean President (Kraus 2017: 261–2).
All the foregoing arguments suggest that states form defensive military alliances as a way to manage a security challenge posed by some external threat. In that sense, Walt’s original theory, at its core, prevails. However, states have sometimes established alliances so as to mitigate conflict with each other. Paul Schroeder (2004: 196) observes “that alliances in practice do not always serve to increase a nation’s power and