Introduction
The COVID-19 pandemic was an unprecedented test of the military’s role during emergencies. Nearly every country in the world was affected by the pandemic between March and July of 2020, and a majority of countries responded in this initial phase by deploying the military. These deployments, however, varied significantly in scope and character. In Israel, military units manufactured ventilators and personal protective equipment; in Italy, soldiers patrolled the streets of major cities to enforce restrictions on movement; and in Morocco, the military announced lockdowns from loudspeakers on tanks. Footnote 1
In this paper, we have two goals. First, we document the variation in early pandemic-related deployments with an original dataset covering every country with a standing military during the initial pandemic response. Our dataset captures seven distinct categories of pandemic-related deployments, listed here from most to least prevalent: logistic operations, enforcement operations, international involvement, border protection, information provision, intelligence operations, and domestic protection. Second, we demonstrate that capacity and governance explain a significant amount of the variation in pandemic-related deployments, shaping the way that countries choose to respond to the pandemic militarily.
Overall, three quarters of countries with a standing military responded with a deployment at the outset of the pandemic. Logistics operations and enforcement of lockdowns, mandates, or restrictions on movement were the most common kinds of deployments. Countries with particularly high numbers of cases were more likely to deploy the military, but the mission of these deployments was shaped by governance. States with elected leaders were more likely to deploy the military in a ‘behind the scenes’ capacity, relying especially on logistic support and border protection. Incumbents facing reelection were particularly sensitive to electoral concerns and avoided domestic enforcement operations especially when the next election was imminent. Footnote 2
Military deployments in crisis situations
Securitization theory argues that when an issue is considered a security threat it is framed as existential and, therefore, critical to survival (Buzan et al., Reference Buzan, Wæver and de Wilde1998). In these situations, the threat is elevated above normal political processes. Audiences empower securitizing actors (like politicians) to take exceptional measures to mitigate these threats, including extraordinary allocations of resources (human, financial, etc.) (Kuehn and Levy, Reference Kuehn and Levy2021).
The COVID-19 pandemic presents an unusual case for securitization. The pandemic emerged as a threat to nearly every country in the world during a short period of time. It quickly became a domestic security issue: securitized language, such as being ‘at war’ with the virus, and the politics of exceptionalism were hallmarks to many countries’ pandemic response (Kirk and McDonald, Reference Kirk and McDonald2021). Indeed, we are not the first to demonstrate that a majority of countries deployed the military in some fashion when responding to the pandemic (Erickson, Kljajíc and Shelef, Reference Erickson, Kljajíc and Shelef2022).
Regulations of types and levels of military deployment are a function of a plethora of institutional, historical idiosyncratic factors that affect the military’s role in a particular country (Passos and Acácio, Reference Passos and Acácio2021; Bove et al., Reference Bove, Rivera and Ruffa2020: 264). Due to their totalitarian past, countries like Germany, Italy or Japan impose severe restrictions in their use of the military, particularly in domestic functions. Yet even those countries utilized the military in the pandemic. Despite cross-national differences in the military’s role in society, we expect that all countries with a military have the potential to mobilize following the shock of the pandemic, but that those mobilizations might take different forms. Footnote 3
Capacity
What shaped military responses to the COVID-19 pandemic? Securitization theory would suggest that, as a first step, state capacity to manage the virus played a role. Securitization is a response to an emergency situation; we therefore expect that the perceived emergency is influenced by the severity of the pandemic and will predict deployments in response to the virus.
Hypothesis 1: All types of deployments are more likely in countries where the pandemic is more severe.
The threat of the pandemic is additionally related to the state’s pre-existing capacity to manage the virus. A country with a poor health care infrastructure faces a greater existential threat from the pandemic.
The relationship between health capacity and military responses to the pandemic is not straightforward; even a country with excellent pre-existing health capacity may view the pandemic as an existential threat and mobilize the military to respond to it. A country with low health capacity, meanwhile, may face a greater threat from the pandemic, but may not have the infrastructure in place for the military to assist with pandemic response.
A crucial factor at play in this context is the popularity of the military. In countries where the military is very popular, its deployment may enhance public support for the elected leaders. Previous research has shown, for instance, that former French President Hollande extensively deployed the military domestically in an antiterrorist function to boost his low public support (Bove et al., Reference Bove, Rivera and Ruffa2020). In these contexts, we expect that public leaders may weigh the benefit of deploying the military domestically against the cost of doing so. Those considerations are specific to each country in terms of political culture and popularity of the military. Footnote 4
We do not predict that pre-existing levels of health capacity will make a state more or less likely to deploy the military in response to the pandemic. Rather, we expect that health capacity will affect the type of mission deployment. Countries with low domestic health capacity will be forced to use the military to shore up deficiencies in domestic infrastructure. Countries with high health capacity, meanwhile, can use the military in other roles.
Hypothesis 2: Countries with a high level of domestic healthcare capacity are more likely to deploy the military in a non-domestic capacity, including international deployments or deployments to protect the border.
Electoral politics
Securitization, as a political process, will unfold differently depending on electoral politics. Kenwick and Maxey (Reference Kenwick and Maxey2022) find that democratically elected leaders use the military in ways that bolster their public approval and improve their chances of political survival. Elected leaders have incentives to avoid unpopular actions and take actions that curry favor with the electorate. We therefore distinguish between ‘boots on the ground’ deployments—which are highly visible and civilian-facing—and ‘behind the scenes’ deployments—which assist with pandemic response in a way that does not create interaction between military personnel and the average citizen. We argue that ‘boots on the ground’ deployments are riskier for democratically elected leaders because they give the impression that the pandemic is out of control. Because using the military directly to manage citizens breaks norms of civilian control, it is likely to lead to negative assessments of leadership. Footnote 5
Hypothesis 3: Countries with elected leaders are more likely to deploy the military in a ‘behind the scenes’ capacity—to provide logistics, border protection, or intelligence—and less likely to deploy the military in a ‘boots on the ground’ capacity—to enforce lockdowns or protect domestic infrastructure.
Even among elected leaders, there is considerable variation in sensitivity to electoral concerns. A leader facing a reelection campaign has more to lose from an unpopular pandemic response than does a leader who does not need to run for reelection in the near future. Pulejo and Querubín (Reference Pulejo and Querubín2021) find evidence of such effects for other kinds of pandemic responses (unrelated to military deployments). We postulate that leaders who are eligible to run for another term and face an election in the near future will be particularly sensitive to electoral concerns.
Hypothesis 4: Countries where incumbent leaders are facing an imminent election are particularly likely to avoid deploying the military in a ‘boots on the ground’ capacity.
Data
We collected original data on pandemic-related military deployments taking place from the beginning of the pandemic through August 2020 in order to focus on states’ initial securitization responses. On a country-by-country basis, we conducted targeted searches for news articles on Google using the following search string: military AND (COVID OR Corona) AND [country name]. Footnote 6 We documented military actions reported in mainstream news media, government sources, or non-governmental organization sources (for example, the World Health Organization). In cases where information was sparse, we supplemented the initial search with additional searches of the Factiva news database.Footnote 7 Media reports and government press statements constituted the majority of our raw material, which consisted of some 1500 separate reports. Footnote 8
Take the case of Sudan, which exemplifies the data landscape. We relied on seven articles to code Sudan as engaging in logistics and enforcement activities. An article from All Africa described the military enforcement of an April 2020 lockdown, relying on statements from various leaders (including a TV appearance by the Minister of Health), interviews of residents, and social media.Footnote 9 Additional news reports from the Agence France Press newswire, the European Council of Foreign Relations, and others, provide additional information on military enforcement and logistics (which involved assisting health authorities in preparing a quarantine center). We choose to highlight Sudan because it is considered to be a ‘closed autocracy’, that is, a country which not only restricts rights but also fails to hold multiparty elections. Footnote 10 In that sense, one might presume that it could be a ‘hard case’ for obtaining information on government military responses to COVID. We did not find, however, that information landscapes varied noticeably across different regime types. This is likely to be the case for several reasons. First, autocracies generally do not have the capacity to enforce complete censorship Footnote 11 ; even a highly capable and well-resourced state like China has struggled to control its citizens’ spread of information and opinions (Chang et al., Reference Chang, Hobbs, Roberts and Steinert-Threlkeld2022). Second, autocracies usually pick and choose which topics they consider to be threatening (King et al., Reference King, Pan and Roberts2013). Autocratic governments would be unlikely to restrict information on their own policy responses for which they sought citizen cooperation. In fact, our source material indicates that government statements were often the primary sources for media reports, as governments—democratic and autocratic alike—sought to inform the populace of the new policy in place, as well as the repercussions which would be incurred for non-compliance. Finally, because even OECD democracies were responding with lockdowns and other restrictive policies, autocracies had no reason to fear that the policy itself would result in international censure. Footnote 12
In order to meet the inclusion criteria for our dataset, a deployment must be explicitly motivated by the COVID-19 pandemic. We do not include deployments motivated by factors other than the pandemic or deployments that were planned or initiated before the onset of the pandemic. In practice, distinguishing pandemic-related deployments from other cases was generally straightforward because of the public nature of large-scale deployments and pandemic responses.
We are aware of one other effort to collect data on military responses to the COVID-19 pandemic: Erickson et al. (Reference Erickson, Kljajíc and Shelef2022) similarly present original data on military deployments. Our approach differs in the way that we disaggregate deployments. Erickson et al. (Reference Erickson, Kljajíc and Shelef2022) disaggregate deployments into logistic, healthcare, coercion, and policymaking. We disaggregate deployments into seven distinct categories. These include international deployments—which Erickson et al. (Reference Erickson, Kljajíc and Shelef2022) explicitly exclude—and border deployments, which may be of particular interest to scholars studying the political implications of pandemics (Kenwick and Simmons, Reference Kenwick and Simmons2020). Footnote 13
Types of deployments
We define the categories of pandemic-related deployments below in decreasing order of prevalence. Table 1 shows the proportion of countries reporting each of these kinds of deployments, and Figure 1 shows their geographic distribution.
Table 1. Summary of COVID-related military action
Figure 1. Global pandemic military deployments through July 2020.
Logistics: We define logistic operations as military involvement in public health/medical logistics relating to COVID-19, including providing human resources support to these activities. Involvement is sufficient; the military may or may not take the lead. Footnote 14 Civilian military personnel, including medical staff and logistics staff, count as ‘military involvement’ for the purposes of this variable. Logistics can include aiding healthcare personnel with testing, constructing new hospitals or field hospitals, manufacturing or distributing medical supplies for domestic use, or military personnel treating patients.
Enforcement: We define enforcement operations as military involvement in the enforcement of domestic restrictions on the freedom of movement motivated with reference to COVID-19, such as quarantines and lockdowns. This can include, for example, clearing shops and public spaces or checking on people’s permission to be out of doors. We consider the military involved in enforcement even if they share enforcement authority with civilian police.
International Involvement: We define international involvement as military involvement outside of the national borders, motivated with reference to COVID-19. This variable includes cases where the military (1) is involved with the repatriation of citizens from foreign countries or (2) is involved in providing assistance to one or more foreign countries in any form, be it material (for example, sending respirators) or human resources (for example, sending military medical doctors) or providing transportation (for example, air lifts).
Border Protection: We define border protection as military deployment to conduct or assist in border protection motivated with reference to COVID-19. Involvement is sufficient; the military need not take the lead. Pandemic-related border protection can involve pandemic-specific duties, like checking test results or travel documents, or it might involve standard border procedures motivated with respect to COVID-19. Footnote 15
Information Provision: We define information provision as the military providing information to the populace on COVID-19-related issues, such as government COVID-19 policies or citizen obligations. This can include, for example, announcing a lockdown using loud speakers on tanks (as in Morocco) or military personnel communicating the gravity of the disease among their home communities (as in India).
Intelligence Operations: We define intelligence operations as military intelligence activities related to COVID-19, including technological development (such as tracking apps or facial recognition software) and the provision of analysis. Examples include preventing or countering the spread of misinformation (as in the UK), developing contact tracing apps for smartphones (as in Switzerland), or ‘cyber patrols’ to monitor social media for misinformation or other COVID-related communication (as in Argentina).
Domestic Protection: We define domestic protection as domestic military deployment to protect critical infrastructure, including security targets, personal protective equipment production sites, etc., motivated with reference to COVID-19.
Our categorization captures seven distinct kinds of deployment. Within these seven types of deployment, one might construct aggregate categories—those that are civilian-facing and grant more power to the military, for example (Gibson-Fall, Reference Gibson-Fall2021). We leave the categories disaggregated for our analysis, but explore alternative codings in online Appendix 2.
Research design
We examine the associations between pandemic-related military deployments and capacity and governance using a series of logistic regressions. In each model, the dependent variable is a pandemic-related military deployment. We model each type of deployment separately. We chose this modeling strategy because the various kinds of deployment are not mutually exclusive; a state can, for example, mobilize the military to assist in logistics, enforce lockdowns, both, or neither. This process is best modeled as a set of parallel decisions to mobilize in various capacities. We additionally include a model where the dependent variable is any kind of military deployment. This variable is coded 1 if the country in question deployed the military for any of the seven kinds of deployment we measured.
We first estimate a series of models predicting pandemic-related military deployments among all countries with a standing military. We then estimate a second series of models among presidential democracies with scheduled elections in order to examine the influence of electoral politics on pandemic-related deployments.
Our modeling strategy is designed to identify associations between governance/capacity and pandemic-related deployments. Our models are not intended to capture every possible factor that might affect the decision to deploy the military. Instead, we include measures of our theoretical concepts of interest (governance and capacity) and potential confounders—variables that might affect both governance/capacity and the decision to deploy the military. Our models are thus designed to estimate unbiased associations between governance/capacity and deployments without overparameterizing the model. Footnote 16
Independent variables
The independent variables in our analysis measure aspects of capacity and governance.
Capacity: We conceptualize capacity as two-sided: a state’s capacity to manage the pandemic increases with the resources available to it to control the virus, and decreases with the severity of the pandemic. We measure pandemic severity using data from the World Health Organization (2020 b) on the cumulative number of COVID cases in the country through July 31, 2020. Footnote 17 COVID cases are logged due to their skewed distribution.
We measure a state’s healthcare capacity as the number of hospital beds per capita at the outset of the pandemic (World Health Organization, 2020 a). We additionally include gross domestic product (GDP) per capita in current USA dollars as a general indicator of a country’s financial resources (The World Bank, 2015).
Governance: We are particularly interested in how electoral politics—particularly the need for leaders to consider the consequences of their actions in future elections—influence the state decision to deploy the military in response to the COVID-19 pandemic. We therefore distinguish between democracies and non-democracies. The variable democracy is a binary variable measuring regime type (Marshall et al., Reference Marshall, Robert Gurr and Jaggers2016). It is coded 1 for democratic nations and 0 for all others. Footnote 18
Within the subset of democratic countries with scheduled elections, we include additional variables to measure variation in electoral pressure. This subset consists of 65 presidential democracies with constitutionally scheduled elections, as defined by Pulejo and Querubín (Reference Pulejo and Querubín2021). Election proximity measures the time until the next scheduled election from the first recorded COVID case in the country. Values of this variable are measured in days and converted to years for ease of interpretation. Values of this variable are multiplied by −1 so that higher values correspond to more proximate elections. A country with an upcoming election in six months, for example, would have an Election Proximity value of −0.5.
Incumbent can run is a binary variable measuring whether the incumbent leader is eligible for reelection. We interact this variable with election proximity to test our expectation that leaders facing electoral pressure are less likely to deploy the military in a ‘boots on the ground’ capacity. Incumbent leaders who are ineligible for reelection, or who will not go up for reelection in the immediate future, are less likely to respond to electoral considerations than those who are eligible and whose elections are imminent.
Additional variables: Perceptions of the military are likely to differ between countries due to cultural and historical differences in the military’s role in society. These differences in military popularity might shape leaders’ decisions to deploy the military—a leader has more to lose and less to gain, for example, by deploying an unpopular military. We therefore include a measure of perceptions of military corruption from the Global Corruption Barometer (Transparency International, 2013). Military corruption is measured as an ordinal scale derived from survey data. Respondents are asked, ‘To what extent do you perceive the following categories in this country to be affected by corruption?’ and then prompted to answer the question with respect to the military. Responses are provided on an ordinal scale where 1 represents ‘not at all corrupt’ and 5 represents ‘extremely corrupt’. Values of this variable are averages of the ordinal responses given by all respondents within a country.
All models additionally include a measure of population density, measured as the total population divided by the land area of the country in square kilometers (The World Bank, 2015). Population density has the potential to affect both the spread of the virus within a country and that country’s military response. It could therefore cause omitted variable bias if excluded from the models. This variable is logged due to its skewed distribution.
Results
The results among all countries are displayed in Table 2. The severity of the pandemic is associated with an increase in most kinds of pandemic-related military deployments. Democratic countries were more likely to deploy the military to protect the border. Additional health capacity in the form of hospital beds per capita was associated with an increase in international deployments and a decrease in deployments for information provision.
Table 2. COVID military action, all countries (logistic regression)
Note: *p < 0.1; **p < 0.05; ***p < 0.01.
Figure 2 shows the relative frequency of each kind of pandemic-related deployment as a percentage of democracies and non-democracies.
Figure 2. Prevalence of pandemic-related military deployments by regime type.
Democracies deployed the military more often than non-democracies for each type of deployment that we measured. This finding is in line with theories in civil-military relations that suggest that leaders in non-democracies are aware of the risks attached to deploying the military domestically because they increase the dependence of leaders on the military (Svolik, Reference Svolik2012). Note, however, that this difference is only statistically significant for border protection when controlling for other covariates. Despite lacking statistical significance when controlling for other covariates, there is a pronounced difference in logistic deployments by regime type; three quarters of democracies deployed the military to conduct logistics operations, compared to fewer than half of non-democracies.
Next, we model the prevalence of pandemic-related deployments among the subset of democracies with scheduled elections. Note that we test only for enforcement operations, logistic support, and any kind of deployment in these models due to the sparse number of occurrences of the less-frequent military deployments in this restricted sample. The results of these models are shown in Table 3.
Table 3. COVID military action, countries with scheduled elections (logistic regression)
Note: *p < 0.1; **p < 0.05; ***p < 0.01.
Among this subset of democratic countries with scheduled elections, pandemic severity was associated with an increase in deployments of all kinds and logistic deployments. Domestic health capacity was also a significant predictor of logistics. Strikingly, the interaction between election proximity and reelection-eligible incumbents was strongly and significantly associated with a reduction in enforcement operations. Footnote 19 We visualize the probability of enforcement deployments as a function of time until the next election for eligible incumbents in Figure 3.
Figure 3. Probability of military enforcement as a function of time until next election.
Note: Predicted probability of a military deployment to enforce lockdowns or restrictions on movement as a function of time until next election. Predicted probabilities are calculated using the enforcement model in Table 3 for a country where the incumbent is eligible to run for reelection. Control variables are held at their mean value.
There is a substantively large reduction in the probability of an enforcement operation as the election draws nearer for leaders who are eligible to run for reelection. Notably, this relationship does not hold for incumbents who are ineligible to run for reelection. This pattern is also not present among logistics deployments. It seems that there was a sharp reduction in enforcement operations among electoral democracies only when the incumbent leader was eligible to run for reelection and the election was temporally proximate.
Conclusions
When COVID-19 hit, most countries in the world made use of their militaries from the outset: 73% of countries deployed the military in some form, demonstrating that countries still consider the military a central instrument for responding to emergencies. 58% of countries deployed the military for logistics and 41% for enforcing lockdowns. The widespread reliance on military deployments conceals extensive variation in the missions and tactics of these deployments. Some kinds of military responses—like intelligence operations or protection of domestic infrastructure—were relatively rare.
Our findings indicate that countries use military deployments to increase domestic capacity in an emergency. The shape of these deployments, however, is largely a function of electoral politics. Democracies were particularly likely to deploy the military in a ‘behind the scenes’ capacity, particularly through deployments to the border. Footnote 20
Further research should continue to scrutinize the factors driving specific deployments as well as the differences we find between democracies and non-democracies. The popularity of the military, or its historical role within the country, is likely to influence deployments (Heinecken, Reference Heinecken, Mahé and Wilén2021; Levy, Reference Levy2021). Footnote 21
The global COVID-19 pandemic is a unique opportunity to observe how various governments respond militarily in crisis situations. We hope that our dataset is a worthwhile contribution to scholars attempting to understand military operations in peacetime and domestic military deployments. We additionally find support for recent work theorizing that civilian political officials use the military in ways that bolster their own political image and chances of electoral survival (Kenwick and Maxey, Reference Kenwick and Maxey2022). Our findings offer evidence from an unprecedented emergency situation about the ways that civilian control shapes military deployment.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1755773923000024.
Acknowledgements
We thank Chiara Tulp for her excellent research assistance during the data collection, Nguyen Ha for her help in the initial data cleaning phase and the editors and anonymous reviewers for their helpful and constructive feedback. All remaining errors are our own. We gratefully acknowledge financial support from the Swedish Research Council (grant number 2017-02139) that made the data collection possible and the Royal Swedish Academy in Letters, History and Antiquities.
Open access
