Domestic Military Deployments in Response to COVID-19

Abstract

Militaries are commonly deployed in response to domestic disasters. However, our understanding of this phenomenon remains incomplete, partly because the particulars of disasters make it hard to generalize about deployments used in response. This article leverages the COVID-19 pandemic’s global reach to systematically evaluate common hypotheses about when and how militaries are used to respond to domestic disasters. It presents original global data about domestic military deployments in pandemic response and uses it to assess common theoretical expectations about what shapes whether and how militaries are used in such contexts. The results suggest that decisions about whether to deploy militaries stem from the securitization of domestic disaster relief rather than being responses to specific disaster-related features, state capacity shortcomings, or other social or political factors, even as some of these elements shaped how militaries were used. The article concludes by outlining some hypotheses for future research about the impact of this securitization on civil–military relations.

In the early spring of 2020, as the number of COVID-19 cases skyrocketed, the Israeli Defense Forces’ Chief of the General Staff called for the military to take over the pandemic response from the Ministry of Health (Limor, 2020). While the overtness of the Israeli military’s initiative was perhaps unusual, Israel’s reliance on the military to respond to the pandemic was not unique. As we show below, nearly every state used its armed forces to respond to the pandemic. In part, this is unsurprising since disaster response is part of the mission portfolio of many contemporary militaries (Bruneau & Matei, 2013; Ratchev & Tagarev, 2018). However, our understanding of when and how militaries are used in such contexts remains incomplete, partly because the variation in the particularities of disasters makes systematic, cross-national, investigations of hypotheses about what drives domestic deployments in disaster response difficult to implement.

This article contributes to filling that gap by leveraging the COVID-19 pandemic’s global reach to systematically evaluate common hypotheses about when and how militaries are used to respond to domestic disasters in a setting that is globally comparable. A significant theoretical and case-based scholarship addressing the use of militaries in disaster relief (e.g., Croissant et al., 2010; Egnell, 2008; Gibson-Fall, 2021; Head & Mann, 2009; Kalkman, 2021; Kapucu, 2011; Kohn, 2003; Laksmana, 2010; Levinson, 2008; Schrader, 1993; Tkacz, 2006; Malešič, 2015; Pion-Berlin, 2016; Pramanik, 2018; Ratchev & Tagarev, 2018; Tagarev & Ratchev, 2018) and research on the securitization of healthcare and disaster response more generally (e.g., Bernard, 2013; Curley & Herington, 2011; Chigudu, 2016; Davies, 2008; Enemark, 2009; Kamradt-Scott & McInnes, 2012; Kelle, 2007; McInnes & Lee, 2006; McInnes & Rushton, 2013; Roemer-Mahler & Elbe, 2016; Oshewolo & Nwozor, 2020; Watterson & Kamradt-Scott, 2016) has generated important hypotheses about what drives domestic deployments in disaster response.¹ These hypotheses tend to highlight the role of disaster-related, state (including military) capacity, and social and political factors in driving the use of militaries in disaster response. By evaluating the purchase of these hypotheses cross-nationally, we answer the continuing calls to complement case studies of civil military relations with systematic, cross-national, investigations (e.g., Bruneau & Matei, 2013; Croissant & Kuehn, 2020; Eschenauer-Engler & Kamerling, 2019).

The use of militaries in response to the COVID-19 pandemic has drawn significant attention. The resulting case studies (e.g., Elran et al., 2020; Pasquier et al., 2021) and small-n comparative (usually regional) analyses (e.g., Gad et al., 2021; Opillard et al., 2020; Passos & Acácio, 2021) highlight the variety of roles militaries played, including providing healthcare, logistical support, and law and order, among other missions. Others (e.g., Gibson-Fall, 2021) have identified different degrees of military involvement in the response based on several noteworthy cases. Still other analyses have focused on the military side of the equation, unpacking why militaries themselves might want to participate in the pandemic response (e.g., Kalkman, 2021).

This article complements this existing work in three related ways. First, going beyond individual case studies or regional comparisons, it presents the first global data and evaluation of the correlates of military responses to the pandemic. This allows us to account for potential regional differences and to draw generalizable conclusions without concern for selection bias. Doing so shows that, while some factors thought to shape whether and how militaries are used in disaster response had an impact, many did not. On balance, the findings are more consistent with the widespread securitization of disaster response driving whether militaries are deployed in response rather than such deployments resulting from other hypothesized factors.

Second, we build on the existing literature by disaggregating military involvement in pandemic response into deployments for coercion, healthcare, logistics, and policymaking. This enables the article to explore variation in both whether and how militaries were deployed. It also allows for a more nuanced assessment of common theoretical expectations about what factors shape different kinds of military responses to domestic disasters. Finally, we present an agenda for examining the potential consequences of the widespread securitization of the pandemic response on civil–military relations. The global data we provide can serve as the baseline data for future research exploring these consequences and other potential implications of the militarized response to the pandemic.

When and How Are Militaries Deployed Domestically in Disaster Relief ?

We begin by summarizing the main theoretical expectations about the factors shaping variation in the use of militaries for domestic disaster response. These factors focus on characteristics of the disaster itself, various aspects of state capacity, and domestic social and political attributes, respectively.²

The first factor is a disaster’s scope. Other things being equal, the greater the scale of the disaster, the more likely it is to overwhelm civilian capacity and to initiate the mobilization of military resources, particularly to provide logistics and boost healthcare capacity (Levinson, 2008; Malešič, 2015; Schrader, 1993). Importantly, since large-scale disasters can also overwhelm a state’s ability to maintain order, the scale of the disaster could also shape the military’s use for coercive purposes. This is exactly the possibility envisaged in the 2007 US Department of Defense Authorization Act, which authorized using the armed forces domestically where civilian authorities are “incapable of maintaining public order” in cases of “natural disaster, epidemic, or other serious public health emergency” (Quoted in Head & Mann, 2009, p. 9).

Various aspects of state capacity are also expected to shape whether and how militaries are deployed domestically in disaster relief (Ratchev & Tagarev, 2018). Three related, but analytically distinct, aspects of capacity are particularly salient in this context: the ability of the military to participate in disaster response, the state’s ability to provide public goods and services (administrative capacity), and the state’s ability to maintain internal order and enforce compliance with the law (coercive capacity).³ While military capacity could theoretically be considered part of either administrative or coercive capacity, we treat it as distinct to enable an evaluation of whether state capacity in disaster response is embodied in military or in civilian institutions.

Military capacity to respond to natural disasters can operate in multiple ways. For instance, as is often the case more generally, it is possible that the greater the investment in an institution (in this case, the military), the more likely it is to have the tools, personnel, and organizational skills needed to respond. It is also possible, however, that variation in military spending might not matter as much if military assets are inherently “dual use” and capability developed to prosecute wars can be repurposed for domestic disaster relief.

Importantly, military capacity in disaster response is not determined solely by budgetary resources. Prior experience with disasters or with supporting civilians in emergencies more generally could also cultivate the social expectation that militaries would help and provide militaries with the skills and capabilities required to do so (Ratchev & Tagarev, 2018). This fosters the theoretical expectation that contexts with frequent or severe natural disasters would be more likely to rely on the military for disaster response.

The likelihood that civilian capacity would be overwhelmed could also affect whether and how militaries are deployed domestically for disaster response. Observers often point to a lack of administrative capacity, especially in under-resourced states, as increasing the likelihood that militaries would be used. Malešič (2015, p. 981), for example, hypothesized that since “developing countries tend to lack comprehensive and stable civilian structures for dealing with disasters… disaster response in these countries often relies almost entirely upon the military as well as on international civilian and military assistance. In developed countries, [in contrast], civilian disaster response structures exist and operate at relatively high levels” and, presumably, it takes more to initiate the domestic deployment of militaries in response to disasters (see also Pion-Berlin, 2016; Michaud et al., 2019). Others agree but note that developed states also often have inadequate civilian capacities, increasing their reliance on the military as well (e.g., Head & Mann, 2009, p. 76).

In this vein, scholars highlight specific deficits in administrative capacity that are especially relevant. For example, Pramanik (2018) concluded that India’s ineffective and inflexible civilian bureaucracy led the state to turn to the military for disaster response. To the extent that this experience is more widely applicable, states with higher bureaucratic quality would be less likely to deploy militaries domestically for disaster response in general, and specifically for service provision, because their civilian bureaucracy can adequately respond. Guided by a similar logic, and especially relevant in the pandemic’s context, militaries may be especially likely to provide medical care to compensate for under-resourced health systems (Gibson-Fall, 2021).

Social and political factors beyond capacity are also widely expected to influence both whether militaries are used to respond to domestic disasters and the missions undertaken within those responses. For example, Ratchev & Tagarev (2018) point to the “maturity of civil society” as shaping deployments in disaster response. Here, the logic is that militaries may be less likely to be deployed domestically where civil society is robust enough to respond to a crisis on its own.⁴

Trust in the military could also shape whether it is deployed in a crisis (Gibson-Fall, 2021; Pramanik, 2018; Stepan, 1986, p. 137). Especially where the public trusts the military more than the government, a crisis can make calling on the more trusted institution to protect the nation more likely. As Mares (2014, p. 93) summarized, “to the degree that the public trusts the military, especially if that trust is greater than that for civilian institutions, the attraction of calling on an ‘impartial’ institution… can be overwhelming in a period of crisis.”

We might also expect different regime types to deploy their militaries domestically in different ways. Established democracies may use the military to provide logistical support to civilian operations but may be less likely to use militaries coercively or in policymaking given the strong normative and often legal constraints on doing so (Fitch, 1989; Head & Mann, 2009). Autocracies, in contrast, may be relatively more likely to deploy their militaries coercively, if only because safeguards against military overreach are comparatively weaker. Yet, there could also be differences between different types of autocracies. Civilian autocracies, for example, may be relatively more wary of using the military for disaster response if they fear that it would enhance the military’s role in society and potentially facilitate future challenges to the regime (Svolik, 2012). Military autocracies, in contrast, are less likely to be concerned about this possibility.

Finally, active (or recent) involvement in conflict could also shape whether the military is used in disaster response. Since military forces are a finite resource, states already deploying their militaries could be less inclined to divert scarce military resources to disaster response (e.g., Andreski quoted in Desch, 2001, p. 18). This appeared to be the case, for example, in Mozambique, where the pandemic’s onset sparked debate about whether and how to divert resources to simultaneously deal with both the insurgency and COVID-19.

The pandemic’s global reach and broadly similar character provides an opportunity to systematically examine the extent to which these hypotheses can explain whether and how militaries are deployed domestically in disaster response.

Data and Research Design

To trace whether and how militaries were used, we searched for reports of their deployment in response to the COVID-19 pandemic in the domestic media of every state with a standing military, in international reporting, in government announcements (including Facebook and Twitter accounts belonging to Defense departments and militaries), and in NGO reports between its onset in the spring of 2020 and August 1, 2020.⁵ We define militaries as including “strategic, land, naval, air, command, administration and support forces” as well as paramilitary forces if these are “trained, structured and equipped to support defense forces and are realistically deployable” (International Institute for Strategic Studies, 2020, p. 520).⁶ We excluded reports of unspecified “security forces” that did not distinguish between civilian police and the military. This likely led to undercounting military deployments. For example, reports from Zambia identified “the police” as engaging in coercive pandemic responses but did not distinguish between the civilian police and the military's “Police Paramilitary Unit.” We also excluded reports that militaries engaged in drills to enforce curfews (as in Mongolia) or were given “emergency powers” (e.g., Egypt) in the absence of evidence of actual deployment. Finally, in states with active civil wars, we only coded the use of the military by the internationally recognized government. This decision could also have undercounted some military responses to the pandemic. In Libya, for example, the rebel Libyan National Army responded to the pandemic, but the military of the internationally recognized government did not appear to do so.

We disaggregate military deployments into four distinct missions that get at the specific expectations noted above: providing healthcare, logistical support, coercion, and pandemic response policymaking. Healthcare and logistics are common (and relatively uncontroversial) aspects of disaster relief operations. We coded militaries as engaging in healthcare if they provided treatment, virus testing, disinfection, or public health education outside of the military.⁷ We coded militaries as providing logistical support if they constructed testing and quarantine sites (distinct from operating them, which counted as healthcare), delivered resources (including food and Personal Protective Equipment (PPE)) to civilians, or repatriated citizens. While we counted military production of PPE as logistical support, we excluded military participation in scientific research leading to disease treatment and the production and procurement of medicines or vaccines. As the American and Portuguese experiences show, militaries played important (sometimes primary) roles in vaccine distribution, but these activities are excluded because they occurred after data collection ended.

The pandemic response also included potentially less consensual elements: coercion and policymaking. Military provision of internal security is especially important since the militarization of public order is often identified as a threat to democratic rule (e.g., Diamint, 2015; Fitch, 1989; Head & Mann, 2009; Stepan, 1986; Kohn, 2003; Mares, 2014; Slater, 2010; Tkacz, 2006). States were coded as using the military coercively if the armed forces enforced curfews or public health regulations, or if they staffed internal checkpoints specifically as part of the pandemic response. For example, the Australian and Czech militaries enforced restrictions on the distance residents could travel. Likewise, in Albania, Brazil, Bulgaria, and Cameroon, among others, militaries patrolled the streets to enforce curfews or restrictions on large gatherings. Overall, this indicator is a conservative measure of military involvement in coercion in disaster response as it focuses on direct domestic pandemic response and excludes both the military’s conventional role at state borders and secondary responses like using the military to repress protests against pandemic restrictions.

Finally, we tracked whether militaries participated in formulating pandemic response policy. We proxy the military’s participation in policymaking by its inclusion in national level task forces with responsibility for setting the overall pandemic policy response.⁸ The nature of these bodies, including the level of their institutionalization, membership, and ways of operation varies significantly across countries (Rajan et al., 2020). Where the membership roster of a task force was not available, we relied on whether someone in a military uniform was pictured in photographs of the task force. This proxy thus includes a variety of military policymaking, ranging from total control through a more subtle power gained by agenda setting and shaping the conversation about what to do (Burk, 2002; Brooks, 2020; Feaver, 2003).

We also include indicators for the pandemic-related, capacity, and contextual factors commonly assumed to shape the domestic use of militaries in disaster response. Pandemic severity is coded in each case three ways based, respectively, on WHO reporting of cumulative cases as of July 31, 2020, cumulative COVID-19 fatalities, and assessments of excess mortality in each country (which may be less subject to state manipulation) (Giattino et al., 2021). The analysis in Table 2 reports results using the first measure (Supplementary Appendix Tables A2–A3 provide results using the alternative measures).

Table 2.

Correlates of Military Deployment in Response to the Pandemic.

	Dependent Variable
	Military Use	Coercion	Logistics	Healthcare	Policymaking
	(1)	(2)	(3)	(4)	(5)
Disaster burden
Pandemic burden	−0.047	0.234∗	0.347∗	0.424∗∗	−0.229∗
	(0.312)	(0.139)	(0.197)	(0.160)	(0.133)
Pre-existing state capacity
Military exp. per capita (logged)	0.635∗	0.315	0.149	−0.178	0.150
	(0.360)	(0.209)	(0.229)	(0.216)	(0.198)
Average number of annual disasters	−0.020	0.005	0.002	0.570∗∗	0.259∗∗
	(0.100)	(0.080)	(0.097)	(0.268)	(0.119)
Bureaucratic quality	0.615	0.099	1.611∗∗∗	0.166	−0.303
	(0.625)	(0.080)	(0.479)	(0.354)	(0.326)
Capacity to provide law and order	−0.156	−0.862∗∗∗	0.246	−0.005	−0.163
	(0.551)	(0.312)	(0.398)	(0.279)	(0.252)
Health equity	−0.486	−0.519∗	−0.735∗∗	0.525	0.126
	(0.460)	(0.279)	(0.328)	(0.287)	(0.251)
Political and social factors
Democracy	2.388	−0.206	1.241	0.012	0.866
	(1.502)	(0.800)	(0.925)	(0.764)	(0.684)
Civil society environment	1.249	−0.520	0.532	−0.273	−0.068
	0.818	(0.373)	(0.502)	(0.385)	(0.343)
Trust gap	0.287	0.529	2.248	3.019∗∗	1.300
	(3.502)	(1.441)	(2.133)	(1.498)	(1.369)
Conflict	0.931	−1.065	2.339∗∗	−0.726	0.205
	(1.024)	(0.698)	(1.035)	(0.781)	(0.649)
Freedom of expression	−2.825	4.070∗	-1.211	−0.631	−2.949∗
	(3.507)	(2.201)	(2.720)	(2.230)	(1.965)
Intercept	−0.836	−1.623	−7.018∗∗	−4.737∗∗	2.746
	(4.503)	(1.896)	(2.950)	(2.192)	(1.822)
Observations	118	118	118	118	118
Log likelihood	−10.795	−54.772	−29.065	−47.902	−63.915
Akaike Inf. Crit.	45.590	133.544	82.129	119.805	151.830

∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Since military spending is a common measure of institutional capacity (Berwick & Christia, 2018; Hanson & Sigman, 2021), our primary indicator of the military’s ability to respond is 2019 per capita military spending (logged) (SIPRI, 2020). Because actual experience can also shape capacity by providing relevant skills and institutional know-how, we also account for the average number of annual disasters a state experienced since 1980 based on the Centre for Research on the Epidemiology of Disasters’ Emergency Events Database (EM-DAT). As robustness tests, the appendix also explores alternative indicators of relevant military capacity, including having used the military to respond to a major domestic disaster since 2000, and the regular use of the military to provide internal security or to support civilians in a crisis.

While some aspects of military capacity are related to coercive and administrative capacity more broadly, we also include different measures of the latter to explore the extent to which state capacity to respond to disasters is embedded in military or civilian institutions. Specifically, we account for the capacity to provide law and order based on assessments of the impartiality of the legal system and popular observance of the law (International Country Risk Guide Researchers, 2018). Prior administrative capacity is assessed using an indicator of bureaucratic quality (relative autonomy from political interference) prior to the pandemic (International Country Risk Guide Researchers, 2018) and, given the pandemic’s medical character, access to healthcare (based on the 2019 Varieties of Democracy (V-Dem) measure of whether high-quality healthcare is provided to all (Coppedge et al., 2020)).⁹

The potential influence of civil society’s ability to cope with a crisis is captured using V-Dem’s 2019 assessment of the extent to which countries have many diverse civil society organizations, and it is considered normal for people to be occasionally active in one of them (v2csprtcpt) (Coppedge et al., 2020).

We code regime type using a binary version of V-Dem’s 2019 Regimes of the World (v2x_regime) variable (Coppedge et al., 2020), in which autocracies include both closed and electoral autocracies, and democracies include both electoral and liberal democracies. Given findings that the choice of regime type data matters (Lueders & Lust, 2018), the supplementary analysis verifies that the main results hold using alternative measures and sources for categorizing regimes (see Supplementary Appendix Tables A4–A6).

Data on the trust gap between the military and civilian institutions is based on Gallup World Poll, World Values Survey, Eurobarometer, and Arab Barometer surveys.¹⁰ The “Trust Gap” variable in Table 2 reflects the difference between the percentage expressing trust in the military and those expressing trust in the government in each state. As robustness tests for institutional trust, the appendix also investigates perceptions of corruption in the public sector and whether states have compulsory military service.¹¹ Finally, conflict engagement is coded as participation in international conflict, civil war, or counterinsurgency in 2019 (Gleditsch et al., 2002; Pettersson & Öberg, 2020).

How States Used Their Militaries Domestically in Response to COVID-19

Almost every state (95%) used their armed forces to respond to the pandemic in some fashion (Table 1). The only states that did not (as of August 1, 2020, when data collection ended) were Belarus, Guyana, Haiti, Libya, Mozambique, Tajikistan, Turkmenistan, and Zambia. Some of these claimed that they had no COVID-19 infections (e.g., Turkmenistan). Others were actively fighting insurgents and likely had little remaining military capacity to respond (e.g., Mozambique, Libya). As Table 1 shows, militaries were commonly used in both conventional disaster relief operations (logistics and healthcare) and in ways that deviate from these strictly humanitarian missions (coercion and policymaking). The near universality of military deployments and the commonality of these missions reflects the extent to which domestic disaster response has been securitized.

Table 1.

Military Deployment in Response to the Pandemic.

Variable	Proportion	SD	N
Military use	0.95	0.21	170
Coercion	0.71	0.45	170
Logistics	0.72	0.45	170
Healthcare	0.66	0.48	170
Policymaking	0.36	0.48	170

Table 2 presents the results of Firth logistic regressions evaluating the factors shaping whether militaries were used in pandemic response (model 1) and how they were used (models 2–5).¹² All models account for potential bias arising from the use of public reporting as the main data source by controlling for a state’s freedom of expression.¹³ Overall, the results suggest that the securitization of disaster response played a bigger role in driving whether militaries were deployed domestically than any particular disaster-related, capacity, or contextual factor. The results do support some of the factors commonly thought to shape how militaries are deployed in domestic disaster response, but show that others have less generalizable purchase.

Despite expectations that disaster severity would shape whether militaries would be deployed, the pandemic’s burden was not significantly related to using the military in response. Many states called on the military preemptively, even before they had a significant number of cases (e.g., Georgia and Uruguay, among others). This finding is consistent with the deep and widespread securitization of disaster response and reinforces the extent to which armed forces often lead such efforts. Notwithstanding the broadly consensual character of such deployments (Croissant et al., 2010; Malešič, 2015; Ratchev & Tagarev, 2018), as we note in the following section, this reality reflects an additional, and potentially self-reinforcing, shift away from Huntington’s (1957) ideal of a military solely devoted to the management of external violence.

The pandemic’s severity did affect how militaries were used, and in the expected direction. The worse the pandemic, the more likely were militaries to be used for logistical support, healthcare, and coercion.¹⁴ This pattern is consistent with the possibility that while the global securitization of disaster response shapes whether militaries are used, objective factors may shape how they are deployed in response.

Military spending shaped both whether militaries were deployed and the uses to which they were put. States that spent more on their militaries per capita in 2019 were more likely to deploy their military and, in some specifications, to use the armed forces for coercion (though the latter results are sensitive to model specification, see appendix). Importantly, per capita military spending still shaped whether militaries were used in pandemic response even when the analysis was limited to cases that are presumed to have sufficient resources to do so (those in the top 75% of military spending) (Supplementary Appendix Table A11). This suggests that the level of investment in the military matters, even among states that have the technical means of responding.

The impact of experience and other ways of capturing relevant military capacity are more mixed. On the one hand, states with more experience with natural disasters were more likely to include the military in pandemic-related policymaking. Table 2 suggests that such states were also more likely to include the military in providing healthcare, though that result is sensitive to alternative ways of capturing a history with disasters (Supplementary Appendix Tables A14–15). These findings are consistent with the intuition that regularly experiencing significant disasters fostered the prior securitization of disaster response and the inclusion of the military in the policymaking process regarding disaster response.

On the other hand, specific experiences using the military to respond to large-scale domestic disasters or to provide internal security or emergency aid to civilians did not systematically shape the likelihood that militaries would be used for coercion or logistical support during the pandemic, where we might have expected this experience to be most impactful (Supplementary Appendix Tables A12–13).¹⁵ In other words, it was not the case that contexts habituated to using the military in domestic disaster relief or in other related ways differed significantly from those without these experiences in responding to the pandemic. While fully disentangling which aspects of military capacity matter requires additional research, overall, the results are more consistent with states turning to the military to recoup the investment in it than with military use being driven by the inherently dual use character of military skills and materiel or with specific relevant experience. As we note below, this is one key reason that the securitization of disaster response could foster a self-reinforcing cycle that would further entrench the military’s role in this arena.

Civilian state capacity did not significantly affect whether militaries were deployed, though it did shape how they were used. Consistent with the intuition that where high-quality healthcare is widely available and militaries are not needed to fill the gap by creating new facilities, states with more equal access to healthcare were less likely to use their military for logistical support. While Table 2 suggests that access to healthcare was negatively related to using the military for coercion, perhaps because the military was not needed to enforce public health measures, that finding is not robust to controlling for region (Supplementary Appendix Table A17). However, states with less prior capacity to provide law and order were significantly and consistently more likely to use the military for coercive tasks during the pandemic.

Our findings, however, also suggest that the story is more nuanced than a simple case of the military stepping in when civilian agencies are incapable of doing the job. Specifically, as Table 2 shows, the expectation that less healthcare capacity would be associated with using the military to provide healthcare during the pandemic is not supported. Likewise, despite the relative lack of healthcare capacity in sub-Saharan Africa (World Risk Report, 2019) and assumptions that African militaries would be especially likely to provide healthcare (Gibson-Fall, 2021), they were not significantly more likely to do so (Supplementary Appendix Table A17). The absence of administrative capacity in a particular domain thus does not appear to be sufficient for states to call in the troops for help in that arena.

Similarly, expectations that low civilian bureaucratic quality would be associated with more likely use of the military were also not supported. In fact, states with relatively autonomous and capable bureaucracies were significantly more likely to use the military to provide logistical support. Together with the impact of military spending, this counterintuitive finding suggests that much of the institutional capacity to respond to disasters is now located in the military rather than in civilian institutions, even in high-capacity states.

Most of the social and political factors hypothesized to shape domestic military deployments in disaster response are also not systematically supported. Challenging the normative expectation that democracies would be less likely to use the military for domestic coercion, fully 70% of democracies did so during the pandemic and they were not significantly different from autocracies in this regard.¹⁶ This is a potentially significant finding given the common idea that using the military to provide internal security poses a threat to democratic rule. Additional research is needed to evaluate whether, as we suggest below, the widespread securitization of disaster relief reflects a subtle shift in the balance of power between military and civilian institutions, or whether the military’s provision of internal security in such contexts is not necessarily detrimental (Pion-Berlin, 2014, 2016). At a minimum, our findings suggest that the images of soldiers patrolling the streets after natural disasters like Hurricane Katrina or the Manitoba floods are more characteristic of how democracies use their militaries domestically than often appreciated.

The supplementary analysis in the appendix shows that, contrary to some expectations, newer democracies do not behave differently than established ones (Supplementary Appendix Table A18). The appendix also explores the possibility that the observed variation in military deployments was shaped by fear that using the military would facilitate greater military influence in politics and society. This logic is especially relevant because the observed similarity across regime types could be driven by the absence of this fear in both democracies (because they have strong civilian oversight) and military dictatorships (because the military is already in control). To the extent that this is the case, we should observe a difference between these regimes and civilian autocracies, where the fear of counter-regime mobilization by the military may be highest.

This logic, however, finds only limited empirical support. As Supplementary Appendix Table A19 shows, civilian dictatorships did not differ in their use of the military for coercion or healthcare, though they were significantly less likely to use the military for logistical support. Supplementary Appendix Table A20 accounts for the potential impact of fear of military intervention using a history of coup attempts, reasoning that such fears would be greater in states with a history of attempted coups. As it shows, a history of coup attempts was not significantly correlated with abstaining from deploying the military in pandemic relief. Similar results are obtained using a history of successful coups (rather than attempts) or the time elapsed since the last coup attempt (not shown).¹⁷ At the very least, these results suggest that the securitization of the response to the pandemic was so widespread that it swamped differences between regime types that may be relevant in other contexts.

The absence of a large, diverse, and active civil society was also not significantly correlated with military deployments in response to the pandemic. Likewise, relative trust in the military was also not significantly correlated with whether militaries were deployed in pandemic response. Table 2 suggests that states with greater trust in the military than in civilian institutions were, however, more likely to use the military to provide healthcare, though this result is not robust to alternative (though less direct) ways of operationalizing trust (Supplementary Appendix Tables A7–A9). The mixed findings on the impact of trust call for a reassessment of the mechanisms through which institutional trust shapes when and why states turn to the military to solve domestic problems.

Finally, despite Mozambique’s experience, we find no evidence of a systematic trade-off between recent engagement in conflict and domestic pandemic deployments. If anything, the observed pattern is more consistent with the intuition that investments in logistical capacity as part of conflict can be repurposed for other uses rather than with a trade-off between disaster relief and combat operations (Bruneau & Croissant, 2019; Desch, 2001; Kohn, 2003; Laksmana, 2010; Oshewolo & Nwozor, 2020). It is still possible that the trade-off works in the opposite direction, and that domestic deployment undermines combat readiness, though assessing this empirically requires still unavailable data about conflict incidence in the pandemic’s shadow.¹⁸

In summary, the findings that disease burden, shortcomings in civilian state capacity, and social and political factors were not systematically related to whether the military was used reflects the importance of conceptions that the military ought to be used in response over and above any objective indicator of the need or ability to do so. Notwithstanding expectations that militaries would step in as a last resort to address specific needs, the pattern of pandemic responses demonstrates that militaries now play a much wider and leading role in domestic disaster response across the globe (Hoffman & Hudson, 2009; Ferris, 2012).

Potential Implications for Civil–Military Relations

To be sure, the extent of securitization can vary across cases (Curley & Herington, 2011) and it is too soon to tell what, if any, effects the widespread securitization of the pandemic response will have on civil–military relations or whether there will be overt harm to civilians or democracies as a result. Our findings suggest, however, that even if the militarized response to the pandemic does not fundamentally restructure civil–military relations it could nonetheless exacerbate the securitization of domestic disaster relief and lead to even greater involvement of the military in such activities in the future. We conclude by outlining several theoretically driven hypotheses about how these dynamics could develop.

First, we can expect militaries themselves to lobby for their involvement in disaster response. As others have noted, “Assisting relief efforts can improve the military’s image… and may also be a way for the military to diversify their role at a time when armed forces throughout the European countries are experiencing budget cuts” (Ratchev & Tagarev, 2018, p. 142. See also Egnell, 2008; Kalkman, 2021; Malešič, 2015). While this assessment was made with an eye to the impact of disaster relief abroad, its logic applies equally to domestic disaster deployments. As military budgets are perennially “on the table,” we can expect militaries to continue advocating for their expansion into this arena, perhaps especially where there is relatively less concern about armed conflict with neighbors.

Second, these dynamics could be reinforced by run-of-the-mill “mission creep” and the military’s bias for action if a new “new professionalism” that includes disaster response and related functions emerges (Brooks, 2016; Pramanik, 2018). In a process analogous to the strategic preparation for war (Golby, 2015), the widespread use of the military for domestic disaster response could change how civilian and military institutions prepare for future disasters, thereby making it more likely that the military would be used. For instance, further investments in the military, especially if made at the expense of civilian disaster response institutions, make it more likely that the military would be called upon in the future for purely practical reasons. Reinforcing this dynamic, successful disaster responses, in turn, because they enable militaries to demonstrate publicly how efficient, capable, and important they are, could also increase the popularity of, and trust in, the military, perhaps especially when counterposed with perceptions of lackluster civilian responses. While we did not find a correlation between trust in the military and its use in response to the pandemic, it is still possible that such increased popularity would lower the barrier for future domestic deployments in disaster relief by reducing potential political costs of doing so.

This increased popularity, as well as the institutional and personal contacts with local leaders and interest groups enabled by domestic deployment in disaster relief (Kalkman & Groenewegen, 2019), can also translate both into more expansive demands for budgets and prerogatives by the military and into a reduced willingness or ability of civilian leaders to reject those demands (Beliakova, 2021; Brooks, 2019; Feaver, 2003; Golby et al., 2017; Ratchev & Tagarev, 2018). Indeed, this dynamic was already evident in recent debates over military spending in the European Union (Lazaro, 2020).

Although civilian leaders may be able to recover some power delegated to the military in a crisis (Croissant et al., 2010; Pion-Berlin, 2014), the political and institutional consequences of militarizing domestic disaster relief may still strengthen the military’s domestic bargaining position and increase institutional friction in ways that render civilian authorities relatively weaker than they were before. For example, shifting institutional responsibility for domestic disaster relief to the military can ultimately constrain civilian oversight by reducing the availability of civilian expertise in disaster response. As has been noted in other contexts, over time, the bureaucratic marginalization of civilian expertise limits the effectiveness of civilian oversight and weakens a critical check on military power (Beliakova, 2021; Cottey et al., 2002; Trinkunas, 2005).

Even if it does not lead to coups, human rights abuses, and other instances of overt harm, we can nonetheless expect the expansion of what militaries see as their appropriate realm of action to increase contestation of the kind observed in Israel over whether militaries or civilian authorities are best positioned to act in response to disasters. Importantly, the existence of this contestation itself and its expansion into a new policy arena marks a shift in the balance between civil and military authorities (Beliakova, 2021; Brooks, 2016; Cottey et al., 2002; Croissant et al., 2010, pp. 957–958; Stepan, 1988).

Indeed, viewed through the lens of “how and how much” influence militaries exert in a particular domain (Croissant et al., 2010), our findings suggest that militaries already play a significant role in the once civilian arena of domestic disaster response. Whether this process of securitization expands, as we suggest it may, the conditions under which it varies, and whether it has negative consequences for civil–military relations, democracy, or other outcomes of interest, necessarily remains a matter for future research.

We hope that such future work would both build on the data we provide as the baseline from which to evaluate these possibilities and expand on its limitations in several ways. First, while we describe important patterns in the correlates of how militaries were used in pandemic response, future research can determine their causal impact more directly. While the (lack of) variation in whether militaries were used is most consistent with the notion that relying on militaries for domestic disaster is driven by the securitization of natural disasters—broadly evident in the common rhetoric of a “war” against the virus—than with other explanations, the content analysis required to provide a direct measure of disaster response securitization is beyond the scope of this article. We hope that future work will develop such a measure to test our main empirical finding more directly.

Second, the ability to draw generalizable conclusions that hold cross-nationally necessarily comes at the cost of understanding particular cases, especially where they deviate from the general trend. We encourage additional research highlighting the processes that lead to the militarization of domestic disaster relief and their consequences in a wide range of contexts. Comparative qualitative research may be especially important, for example, in fleshing out the dynamics of military participation in pandemic policymaking. Doing so may capture important nuance that cannot be detected by our relatively coarse indicators. Such studies would thereby provide greater clarity about the mechanisms that we discuss but do not test directly.

Third, our indicators do not track the balance of activities undertaken within the four broad categories we use or the extent to which these activities were militarized. The latter may be especially important for civil–military relations as the limited use of the military for a specific activity may have different implications than using the military in a leading role. Future research could thus further disaggregate our indicators and develop criteria for measuring the extent of militarization in ordinal or interval terms (Passos & Acácio, 2021; Gibson-Fall, 2021). Such future analyses could also track finer grained aspects of military deployments including who led the response, the length of operations, and whether they formally ended.

Fourth, as noted above, additional factors could also shape domestic deployment in disaster relief, prominently including the politicization of disasters and variation in militaries’ organizational culture. Future work could develop cross-national indicators that would enable the systematic assessment of these factors. Finally, future research could also address other pandemic-related tasks in which the military was involved, including border security and the production and distribution of vaccines, among others, that were excluded from our analysis.

In conclusion, the pandemic’s global reach provided an opportunity for the systematic evaluation of common assumptions about what shapes the use of militaries for domestic disaster relief. The results reveal that objective and contextual factors are not systematically related to whether such deployments occur. Instead, the empirical pattern is more consistent with the claim that disaster response around the world has been securitized to such an extent that militaries are the first, rather than last, resort in domestic disaster response.

Author Biographies

Peter Erickson is an active-duty army officer and a PhD candidate in Political Science at the University of Wisconsin-Madison. His research explores the impacts of political polarization and military prestige on the political involvement of the US military.

Marko Kljajić is a PhD student at the University of Wisconsin-Madison. His research examines the politics of post-conflict reconciliation with a focus on how common institutions can bridge deep social divisions in the aftermath of conflict.

Nadav Shelef is Professor of Political Science at the University of Wisconsin, Madison. His research focuses on nationalism, territorial conflict, and Israeli politics. His most recent book is: Homelands: Shifting Borders and Territorial Disputes (Cornell University Press, 2020).

ORCID iD

Nadav Shelef https://orcid.org/0000-0002-2037-9818