The WHO and Border Restrictions

The WHO’s authority to provide guidance on border restrictions comes from the IHR. The regulations, which date back to the 1850s Sanitary Conferences, have always had the dual purpose to ‘provide a public health response to the international spread of disease’ while avoiding ‘unnecessary interference with international traffic and trade’ (World Health Organization 2005, 1). Formalized by WHO member states in 1951 as the International Sanitary Regulations, they were renamed the International Health Regulations in 1969. Motivated by SARS, states revised the IHR in 2005 to address longstanding issues, including the tendency of states to impose border restrictions during outbreaks that had little public health benefit (Davies, Kamradt-Scott, and Rushton Reference Davies, Kamradt-Scott and Rushton2015).Footnote ¹

Under the current IHR, when the WHO Director-General declares a public health emergency, they are supposed to make Temporary Recommendations about whether (and which) border measures are warranted to respond to a given outbreak. Though the WHO could recommend the adoption of border measures, it most often recommends against ‘any trade or travel restriction’, or some variation of that language (Worsnop, Nass, et al. Reference Worsnop, Nass, Grépin and Lee2023). The WHO has long held the view that most border measures cause more harms than public health benefits.

The WHO’s Temporary Recommendations under the IHR are not binding. But, states that do not abide by them are supposed to follow a process outlined in IHR Article 43. States can impose so-called ‘additional health measures’, including international travel and trade restrictions that differ from what the WHO recommends as long as measures are not ‘more restrictive of international traffic and not more invasive or intrusive to persons than reasonably available alternatives that would achieve the appropriate level of health protection’; measures are based on science and guidance from the WHO; and, states provide their public health rationale to the WHO for measures that significantly interfere with international traffic, notify the WHO of measures within 48 hours, and review measures at least every three months (World Health Organization 2005, Article 43).

As Worsnop et al. (Reference Worsnop, Grépin, Lee and Marion2023) point out, the structure of Article 43 gives states flexibility in meeting their obligations. As in other international agreements, this flexibility may have encouraged states to sign the agreement in the first place. Such flexibility also means it is difficult to identify legal compliance or non-compliance with IHR Article 43. When states do not follow WHO advice, compliance or non-compliance is determined by whether a state follows the above-described process. An important question is whether a state provides sufficient justification to the WHO for its policy. Neither state justifications nor the WHO’s responses are publicly accessible. Therefore, while most states were likely in legal non-compliance during COVID-19 because most failed to notify the WHO of their border measures even though Article 43 requires notification (World Health Organization 2020c; see also, 2020a), no information is available on which states notified or whether they provided sufficient justification. This article is not interested in analyzing legal compliance, but, rather, in examining whether and why states adopted border measures at odds with WHO guidance, which is supposed to reflect current evidence on public health interventions. As such, the article uses the phrase ‘border restrictions’ to refer to border measures inconsistent with WHO recommendations, which is fully defined in the data and methodology section below.

Given this context, from the start of COVID-19 in January 2020, the WHO advised against trade or travel restrictions. When the WHO declared COVID-19 a PHEIC on January 30, it issued that advice as a Temporary Recommendation, noting that the WHO does ‘not recommend any travel or trade restriction based on the current information available’ (World Health Organization 2020b). The WHO’s Director-General further explained on Twitter that ‘[the] WHO doesn’t recommend limiting trade & movement. Travel restrictions can cause more harm than good by hindering info-sharing & medical supply chains & harming economies. We urge countries & companies to make evidence-based, consistent decisions’ (Ghebreyesus Reference Ghebreyesus2020). Yet, close to 65 per cent of states imposed border restrictions by the end of February, and all did so by the end of March 2020.

Existing Explanations for Why States Disregard WHO Guidance

Why did states impose border measures at odds with WHO guidance? And why did some states impose such measures more quickly than others? Existing scholarship on international cooperation and the use of border measures during outbreaks points to international-level consequences, state (in)capacity, and domestic political factors as possible explanations. For one thing, fear of consequences from the WHO was likely not weighing heavily for most states. The WHO has limited authority to enforce state obligations under Article 43. The WHO can request information and justifications from states about their border measures during outbreaks. The WHO can also evaluate whether states’ justifications are sufficient. Lastly, the WHO can track states’ border measures and ‘name and shame’ those that do not follow Article 43. But the organization inconsistently uses the first two strategies and has avoided the third (Kamradt-Scott Reference Kamradt-Scott2016). For instance, during COVID-19, the WHO was notified about only a subset of border restrictions and only some of those included a justification. It is unclear if the WHO evaluated justifications or followed up when states failed to provide justification or notification (World Health Organization 2022a, 4).

The WHO’s hesitance to use its enforcement tools is likely related to its dependence on states for material support. May 2022 was the first time in decades that states agreed to increase the mandatory assessment rates determining the amount of money states contribute to the WHO’s regular budget (World Health Organization 2022b). Difficulties interpreting ‘compliance’ and ‘non-compliance’ could also play a role in why the WHO tends not to ‘name and shame’. Furthermore, the IHR Secretariat has a minimal staff; despite their dedication and hard work, capacity limitations make the above tasks challenging. As a result, states that disregard the WHO’s recommendations face few costs from the organization. In fact, Worsnop et al. (Reference Worsnop, Grépin, Lee and Marion2023) show that due to limited enforcement mechanisms to encourage states to follow WHO recommendations, PHEIC declarations that also recommend against restrictions actually increase the number of states imposing border restrictions.

States have faced few other international-level consequences for not following WHO guidance. While the US warned states to remove bans on US pork imports during the 2009 H1N1 pandemic, there is little evidence that it took retaliatory action (see, for example, US Department of State 2009). Along these lines, China’s status as the second-largest economy with global trade and investment ties did not stop most states from targeting China with border restrictions during COVID-19. Weak international-level material enforcement points to other drivers of variation in states’ border restrictions.

One obvious possibility is that states impose border restrictions because they think the measures have public health benefits, despite the WHO’s guidance. At the outset of the outbreak in early 2020, the WHO’s recommendation against ‘any travel or trade restriction’ reflected an understanding that most border measures used by states during outbreaks cause more harm than public health benefit. COVID-19 added some nuance to that view – the public health impact of international travel measures is actually dependent on conditions like timing, type, and target(s), along with the status of domestic transmission and domestic response measures (Grépin, Aston, and Burns Reference Grépin, Aston and Burns2023; Grépin et al. Reference Grépin, Ho, Liu, Marion, Piper, Worsnop and Lee2021). Indeed, even research prior to COVID-19 showed that, while border measures have not stopped the spread of other outbreaks like SARS and influenza, they can delay spread across borders (Cooper et al. Reference Cooper, Pitman, John Edmunds and Gay2006; Poletto et al. Reference Poletto, Gomes, Pastore y Piontti, Rossi, Bioglio, Chao, Longini, Halloran, Colizza and Vespignani2014; Colizza et al. Reference Colizza, Barrat, Barthelemy, Valleron and Vespignani2007; Cowling et al. Reference Cowling, Lau, Wu, Wong, Fang, Riley and Nishiura2010; Selvey, Antão, and Hall Reference Selvey, Antão and Hall2015). But, most international travel measures actually used by countries had little impact on the spread of SARS-CoV-2 because they did not meet the above conditions (Shiraef et al. Reference Shiraef, Friesen, Feddern and Weiss2022).

But, it is still possible, and even likely, that governments thought they were acting in the interest of public health. Of the states that provided justifications for their border restrictions to the WHO during COVID-19, many cited weak domestic capacity, claiming that fears that their domestic health system could not adequately respond to an outbreak necessitated their reliance on border restrictions (World Health Organization 2022a, 4). And, some analysts questioned the way in which evidence of the utility of international travel restrictions was applied in the early days of COVID-19 (Stanhope and Weinstein Reference Stanhope and Weinstein2020), raising doubt about the WHO’s initial recommendation against the use of these measures. It is also possible that other national health agencies (like the US CDC, China’s CDC, or the Public Health Agency of Canada) offered advice to national governments that differed from the WHO (Leiva Van De Maele et al. Reference Leiva Van De Maele, Kirk, Howard, Verroya and Davies2024).

Reflecting growing uncertainty about the utility of border restrictions, and states’ widespread adoption of measures inconsistent with the WHO’s initial guidance, the organization did update its Temporary Recommendations on February 29, 2020. The WHO noted that some international travel restrictions could be ‘temporarily’ useful under some conditions, such as in ‘settings with few international connections’ and ‘limited response capacities’ (World Health Organization 2020c). The vague language is understandable given the uncertainty associated with the early days of COVID-19. As a result, though, it became difficult to tell which measures would be consistent with this guidance or not. As I describe in the Methodology section below, this is the reason why the primary empirical analysis in this article focuses on the ‘early phase’ of the outbreak before the WHO’s recommendations changed. But, given the shifting evidence base around international travel restrictions and the uncertainty associated with the outbreak of a novel infectious disease, it is possible that even during the early phase of COVID-19 before the WHO recommendations changed, governments thought they were acting in the interest of public health.

There is some empirical support that border restrictions could have resulted from real or perceived vulnerability to COVID-19 spread. For instance, Bickley et al. (Reference Bickley, Chan, Skali, Stadelmann and Torgler2021) find that during COVID-19, more globalized states were more likely to impose border restrictions but that a higher level of government effectiveness (which could proxy for state capacity or confidence in outbreak response) attenuated this relationship. In an analysis of border restrictions during the 2009 H1N1 pandemic, Worsnop (Reference Worsnop2017b) finds that states with weak domestic health infrastructure were more likely than others to ignore WHO guidance, suggesting that this could have also played a role during COVID-19.

There is also evidence that non-public health factors were influential. The association in Worsnop (Reference Worsnop2017b) between weak domestic health infrastructure and border restrictions is pronounced in states with strong democratic institutions that are particularly reliant on maintaining popular support. Other studies also point to non-public health-related factors in explaining border restrictions. In a study of possible epidemiological, economic, and political determinants of travel restrictions between European states during the second wave of COVID-19, Neumayer et al. (Reference Neumayer, Plümper and Shaikh2021) find support for all three. Further, in their conceptual piece on COVID-19 border restrictions, Hoffman, Weldon, and Habibi (Reference Hoffman, Weldon and Habibi2022) suggest that governments were politically motivated. And, Kenwick and Simmons (Reference Kenwick and Simmons2020) argue that increasing investments in border security across states may have played a role in the average stringency of governments’ border restrictions during the first six months of the COVID-19 pandemic. Therefore, while some governments may have been driven by public health concerns, research suggests that other factors like political or economic incentives also drive states’ use of border restrictions.

By identifying the role of factors beyond public health concerns, these studies raise important theoretical and empirical questions. From a theoretical perspective, this work has not rigorously considered how outbreaks trigger fear of foreign ‘others’ and how governments might use these border measures ‘insincerely’. While existing studies do identify non-public health determinants of border restrictions, most have not recognized security theatre as an important explanation (for example, Neumayer, Plümper, and Shaikh Reference Neumayer, Plümper and Shaikh2021; Bickley et al. Reference Bickley, Chan, Skali, Stadelmann and Torgler2021). Some point to related concepts but do not develop a theory connecting security theatre to border restrictions. For instance, Kenwick and Simmons (Reference Kenwick and Simmons2020) suggest that scapegoating and a desire to characterize disease as a foreign threat make border restrictions attractive relative to domestic control measures, but do not differentiate those motivations from the primary argument that ‘preexisting routines of border governance’ influence states’ use of international border restrictions during COVID-19. Similarly, Hoffman, Weldon, and Habibi (Reference Hoffman, Weldon and Habibi2022) mention diverting blame and political theatre as a possible explanation for governments’ international border restrictions, but present it as one of several possible political logics at work. This article systematically lays out how international border restrictions during outbreaks could be useful as security theatre and which governments were likely to be particularly attracted to them during COVID-19.

After carefully laying out the observable implications of these security theatre dynamics, I am able to design an empirical strategy, including using novel data, that is well-suited to evaluate the logic. The article examines all states’ first border restrictions during the early phase of COVID-19. As I describe more fully below, the security theatre logic – though present throughout the pandemic – should be particularly pressing at the start of outbreak. Furthermore, the early phase of the pandemic was fundamental. Not only was this time period the most direct test of governments’ adherence to WHO’s guidance (further discussed below) but these early responses had an outsized impact on the trajectory of the pandemic domestically and globally. Since I focused on this discreet time period, I was able to collect original data for the analysis that reliably codes the date and type of all states’ first international border restrictions. There are existing data sources that include international travel restrictions such as the WHO’s Public Health and Social Measures (PHSM) dataset and the Oxford COVID-19 Government Response Tracker (for example, the latter is used in Kenwick and Simmons Reference Kenwick and Simmons2020; Bickley et al. Reference Bickley, Chan, Skali, Stadelmann and Torgler2021). These datasets are valuable, especially because they cover such a wide range of policy responses during the pandemic beyond international border restrictions. But, media and government (the main sources used in existing datasets) lacked a standardized terminology for describing which measures were used and often misreported when measures were announced or implemented (Lee et al. Reference Lee, Grépin, Worsnop, Marion, Piper and Song2021; Grépin et al. Reference Grépin, Song, Piper, Worsnop and Lee2024). Since this article’s analysis – and the security theatre argument – focuses on timing, the data need careful validation, which I was able to do for all states’ first border restrictions.Footnote ² The following section lays out the argument and hypotheses before turning to the analysis and results.

International Border Measures as Global Health Security Theatre

The article’s main proposition – that many governments use international border restrictions as global health security theatre – reflects the broader literature on border security, which finds that border walls and fortifications are often enacted for reasons outside of ‘hard’ security concerns (Hassner and Wittenberg Reference Hassner and Wittenberg2015; Carter and Poast Reference Carter and Poast2017). While ‘border security’ is often justified in terms of providing actual security from attack, then, its real purpose may actually lie elsewhere. I argue that this is the case for border restrictions during outbreaks.

In his work coining the phrase security theatre, Schneier (Reference Schneier2003, 9) writes that ‘security is both a feeling and a reality’. In the best case, security measures provide both. This concept travels to outbreak response. On the one hand, international border restrictions could be used by governments to, as described above, provide actual security from an infectious disease outbreak. On the other hand, these measures could be used by governments to provide a feeling of security, or even to achieve some other desired policy goal. Schneier offers an example: identification checks at airports, he writes, do not actually provide substantial protection from terrorism. But, airlines supported the ‘security’ measure because they could be seen as doing something about terrorism while allowing them to solve their business issue of people reselling non-refundable tickets (Schneier Reference Schneier2003, 204). Like any countermeasure, then, international border restrictions can be politically useful by promoting a feeling of security and allowing the government to say it is doing something to protect the population. And, as in the airport identification check example, international border restrictions are particularly politically useful when they align with some other policy preference (Friedman Reference Friedman2011, 106).

In other words, when considering how to respond to a given threat, the decision maker’s ‘agenda is about more than security, and often non-security concerns trump security’ (Schneier Reference Schneier2003, 38). In the case of outbreak response, governments have reason to consider more than just providing actual public health protection. Research shows that outbreaks, including the 1918 Influenza Pandemic, SARS in 2003, H1N1 in 2009, Ebola in 2014, and Zika Virus in 2015, have had ramifications for governments’ political support (Boas and Hidalgo Reference Boas and Hidalgo2019; Beall, Hofer, and Schaller Reference Beall, Hofer and Schaller2016; Campante, Depetris-Chauvin, and Durante Reference Campante, Depetris-Chauvin and Durante2020; Walden and Zhukov Reference Walden and Zhukov2021; Sang-hun Reference Sang-hun2015; Gutiérrez, Meriläinen, and Rubli Reference Gutiérrez, Meriläinen and Rubli2022; Al-bakri Nyei Reference Al-bakri Nyei2016).

In the face of an outbreak that could (or is perceived to) cause widespread sickness, disability, and/or fatalities, international border restrictions are an observable policy action that governments can point to as evidence of a ‘strong’ response (see Mesquita Reference Mesquita2007 for a discussion of the political utility of observable measures). Indeed, public support for restrictive policies increases during outbreaks (especially at the outset) because of heightened fear and anxiety (Albertson and Gadarian Reference Albertson and Gadarian2015). The weak evidence base behind most border restrictions during outbreaks, not to mention states’ commitments under the IHR to follow WHO recommendations, is likely not common knowledge amongst the public. Without that context, it seems logical that border restrictions would help to stop the outbreak’s spread (Kobayashi et al. Reference Kobayashi, Cilizoglu, Heinrich and Christiansen2023). As such, border restrictions ‘can make the population feel like it is being protected and provide assurance that the government is doing all that it can, regardless of whether barriers actually prevent disease spread’ (Worsnop Reference Worsnop2017b, 373).

International border restrictions, then, are particularly useful as security theatre because they are visible policy actions but also, in associating the disease threat with foreigners, absolve domestic policymakers while making domestic populations feel safer because they underestimate their risk of infection. Like other types of ‘security theatre’, border restrictions can make the public feel that something is being done (Friedman Reference Friedman2011, 104), whereas not doing something at the border – especially during an outbreak when xenophobia is likely to be activated – could leave the government vulnerable to criticism.

Crucially, border restrictions associate disease spread with arriving foreigners at a time when the ‘othering’ of foreigners is already activated (Dionne and Turkmen Reference Dionne and Turkmen2020). ‘Othering’ – when one group (often a majority or those with access to power) treats marginalized, less powerful, or groups otherwise perceived to be different, as threatening or as if there is something wrong with them – is particularly likely during periods of increased stress or perceived threat, such as a pandemic (Riek, Mania, and Gaertner Reference Riek, Mania and Gaertner2006). Indeed, during outbreaks of disease, ranging from smallpox, bubonic plague and the 1918 influenza pandemic to HIV/AIDS, SARS, and Ebola, ‘migrants and other marginal groups have historically been targets of blame and scapegoating’ (Dionne and Turkmen Reference Dionne and Turkmen2020, E215; see also Dionne and Seay Reference Dionne and Seay2015; Silva et al. Reference Silva, Cade, Figueiras, Roque, Herdeiro and Devakumar2022). COVID-19 was no exception. Reny and Barreto (Reference Reny and Barreto2022) argue that COVID-19, and elite use of racialized rhetoric about the disease in the US, activated xenophobic attitudes, behaviours, and policy preferences. During COVID-19, discrimination and physical violence were often directed at those of Asian descent (or perceived to be of Asian descent). But, Dionne and Turkmen (Reference Dionne and Turkmen2020, E213) note that ‘there are reports of a broad range of people who experienced discrimination and feared stigmatization during the COVID-19 pandemic’. And, this has been a global phenomenon rather than isolated to North America or Europe (Xun and Gilman Reference Xun and Gilman2021).

The tendency to ‘other’ during outbreaks makes border restrictions particularly useful as security theatre. Such measures will resonate with and reinforce the ‘othering’ attitudes that are already present during an outbreak (Adida, Dionne, and R. Platas Reference Adida, Dionne and Platas2020 make a similar point about how politicized rhetoric during the 2014 Ebola outbreak contributed to exclusionary attitudes towards immigrants). This can relieve pressure on policymakers to take steps to address domestic spread, which can be difficult and costly. When considering border restrictions at odds with WHO guidance, then, governments are likely to consider the security theatre benefits.

But, border restrictions also come with a range of short and long-term costs as noted above. Such measures can cause social and economic harms; threaten geopolitical interests and reputation, which can in turn have domestic political consequences; provide a false sense of security and distract from needed domestic control measures; and increase states’ vulnerability to future outbreaks by making it difficult for health workers and resources to get to the source of the outbreak, incentivizing outbreak concealment, and weakening WHO.

Certain governments are therefore more likely to value the security theatre benefits of international border restrictions compared to the costs. I offer one possible operationalization here. If the security theatre logic is operating, then nationalist regimes should be particularly attracted to international border restrictions. Nationalism – an ideology that values membership in a nation over other identity groups like gender, socioeconomic status, political ideology, or religion (though the latter two sometimes intersect with nationalism) – takes a variety of forms (Van Bavel et al. Reference Van Bavel, Cichocka, Capraro, Sjåstad, Nezlek, Pavlović and Alfano2022). When I use the term nationalist, I am referring to what has been called ‘exclusionary nationalism’ (Bieber Reference Bieber2020, 15) or ‘ethno-nationalism’ (Elias et al. Reference Elias, Ben, Mansouri and Paradies2021) and reflects closely Cass Mudde’s (2007) concept of ‘nativism’, an ideology ‘that states should be inhabited exclusively by members of the native group and that nonnative elements are fundamentally threatening to the homogenous nation-state’. Nationalism scholars point out that, even for exclusionary nationalism, membership in ‘the nation’ is not fixed and contestation over membership may be a consistent source of political cleavage or disagreement. Still, most nationalist ideologies ‘define themselves in opposition to an “other” or outgroup’ (Givens and Mistur Reference Givens and Mistur2021, 215). Often, this outgroup includes foreigners and migrants, or those who are perceived to be in such groups. Indeed, nationalist regimes often favour restrictive immigration policies in general (Ko and Choi Reference Ko and Choi2022). Furthermore, exclusionary nationalist regimes have a history of associating foreign ‘others’ with disease threats (Bieber Reference Bieber2022, 18).

As a result, these governments are primed to perceive the security theatre benefits of international border restrictions. Nationalist regimes are likely to see international border restrictions as a way to provide a feeling of security while focusing on the threat of the pandemic as coming from foreigners. This is because these measures offer nationalist regimes a visible policy action that aligns with their underlying ideology, achieves another desired policy goal of further restricting immigration, and may also resonate with, and reassure, these governments’ key constituents (whether those are some portion of the general public in democracies or other key supporters in non-democracies). This dynamic may have played a particularly significant role during COVID-19 because nationalist rhetoric and policy have become ‘more prevalent in global politics in recent years’ (Bieber Reference Bieber2018, 519). Importantly, this recent increase in nationalist rhetoric and policy is not confined to authoritarian regimes but has been observed across political systems including democracies (Givens and Mistur Reference Givens and Mistur2021; Bieber 2018; Reference Bieber2022; Jenne Reference Jenne2018).

This discussion leads to the following general hypothesis:

Nationalist regimes should also impose border restrictions more quickly than others. These states will reach for international border restrictions sooner than others to send a message that they are doing something to prevent the outbreak from crossing the border and to place blame on foreign ‘others’. All else equal, non-nationalist regimes anticipate fewer relative security theatre benefits of border restrictions and can instead wait and see if domestic pressures for restrictions materialize if the trajectory or scope of the outbreak changes, or if the costs and benefits of border restrictions shift in some other way over time.

Finally, if the security theatre logic is operating and governments prefer to ‘enhance … [the] feeling of security’ (Schneier Reference Schneier2003, 10) in order to relieve the pressure to launch a meaningful response, then border restrictions should substitute for domestic measures. Research shows that border restrictions are most effective when part of a comprehensive outbreak response plan that includes domestic measures. This is similar to what Schneier (Reference Schneier2003, 105) refers to as ‘defense in depth’ or what, during COVID-19, was referred to as the ‘Swiss cheese model’ of pandemic response – multiple, layered, countermeasures that provide more protection than one countermeasure alone (Roberts Reference Roberts2020). Conversely, the security theatre logic expects nationalist regimes to use border restrictions in place of recommended domestic measures.

Data and Methodology

I construct an original dataset of the timing and type of each state’s first border measure inconsistent with WHO recommendations, called ‘border restrictions’, during COVID-19. The analysis includes the 196 IHR States Parties, which contain all United Nations Member States.Footnote ³ Below, I describe the data and methodology used in the analysis.

Analysis and Results

The logit models in Table 1 examine whether nationalist regimes are more likely to impose border restrictions in the early phase of COVID-19 (Hypothesis 1) between January 1, 2020 (the day after the first case was reported) and February 29. Model 1 controls for several measures of perceived vulnerability to COVID-19 spread, including GDP per capita, GHSI score, government effectiveness, and distance from Wuhan, China. Note that if perceived vulnerability to COVID-19 is driving state behaviour then we would expect to see a negative relationship between these variables and the likelihood of imposing border restrictions. Model 2 includes additional controls accounting for vulnerability to COVID-19 (population, air passenger volume [if vulnerability to COVID-19 is driving behaviour we would expect a positive association between these variables and border restrictions]), other domestic political and economic factors (the strength of democratic institutions and whether there is a populist in power, the strength of domestic rule of law, the size of the travel and tourism sector, and the level of globalization), and international level factors (participation in China’s Belt and Road Initiative and regional controls).

Table 1. Logit models explaining whether states imposed border restrictions during the early phase of COVID-19

***p < 0.001; **p < 0.01; *p < 0.05; · p < 0.1. Standard errors in parentheses.

Models 1 and 2 show that, all else equal, more nationalist governments are more likely to impose border restrictions (p < 0.05). This finding is robust to a range of alternative explanations. Interestingly, the results show a mixed picture of the role of perceived risk of or vulnerability to COVID-19. While states further from Wuhan, China were less likely to impose border restrictions in Model 1, the association is not statistically significant in Model 2. And, neither GHSI scores nor government effectiveness was associated with state behaviour. Further, states with higher GDP per capita were more likely to impose border restrictions, which is the opposite of what one would have expected if weak state capacity (and perceived vulnerability to COVID-19) were leading states to compensate with border restrictions.

The findings about nationalist regimes are also substantively meaningful. All else equal, a more nationalist regime is 18.5 percentage points more likely than a less nationalist regime to impose border restrictions in the early phase (p < 0.05).Footnote ⁶ Throughout, ‘more nationalist’ and ‘less nationalist’ refer to the third and first quartiles. Figure 1 illustrates the findings. Based on Model 2, the figure shows that the simulated predicted probability of imposing border restrictions increases as the level of nationalism increases.

Figure 1. Simulated predicted probabilities of a state imposing its first border restriction during the early phase of COVID-19 as nationalism increases.

This association holds even when controlling for a range of alternative explanations. As mentioned, states closer to Wuhan, China were more likely to impose restrictions in one model. But, high air passenger volumes were negatively associated with border restrictions, which suggests these states might have been worried about economic repercussions. This possible interpretation is also supported by the negative association between globalization and border restrictions. Other controls, including population, the strength of democratic institutions, populism, size of the travel and tourism industry, domestic rule of law, and participation in the Belt and Road initiative were not significantly associated with state behaviour.

The Timing of Border Restrictions

If governments consider the security theatre benefits of border restrictions, nationalist regimes should not only be generally more likely to impose their first restriction in the early phase; but also, they should do so more quickly than others (hypothesis 2).

Table 2 displays the results of four Cox proportional hazard models. Models 3 and 4 use cross-sectional data to examine variation in how quickly states imposed border restrictions during the early phase of COVID-19. Models 5 and 6 use time-series cross-sectional data with the country-day as the unit of analysis to include two time-varying covariates: for each day of the observation period, the proportion of other states in each region imposing barriers and the number of reported cases in each state. For all models, the observation period begins on January 1, 2020, the day after the first reported case, and ends on February 29. Results in Models 3–6 are reported as hazard rates. Hazards rates can be interpreted as the impact of a given variable on the rate of imposing restrictions relative to a baseline hazard of 1.00. Coefficients greater than one proportionately increase the rate while coefficients less than one proportionately reduce the rate. In Models 5 and 6, standard errors are clustered by state.

Table 2. Cox proportional hazard models explaining the timing of states’ first border restrictions during the early phase of COVID-19 (presented as hazard rates)

***p < 0.001; **p < 0.01; *p < 0.05; ·p < 0.1. Standard errors in parentheses, clustered by state in Models 5 and 6. Note that standard errors are of coefficient values, not of the hazard rates (exponentiated coefficients).

All four models show that more nationalist regimes impose border restrictions more quickly than others. Models 4 and 6 show that this association remains even after accounting for key alternative explanations. That states with more nationalist regimes impose restrictions more quickly offers additional support to the security theatre argument.

The relationship is substantively significant. A state with a more nationalist regime is 2.3 times more likely than a state with a less nationalist regime to impose restrictions on any given day during the early phase of COVID-19.Footnote ⁷ Again, results are mixed for variables measuring perceived risk of or vulnerability to COVID-19 spread. While states further from Wuhan, China impose border restrictions more slowly than others and reported cases are positively associated with the timing of border restrictions (Model 6), GHSI score, GDP per capita, population, and air passenger volume are not significantly associated with the timing of border restrictions. And, government effectiveness is associated with an increase in the hazard rate, which is the opposite of what would be expected if states were driven by weak state capacity and perceived vulnerability to COVID-19. Globalization reduces the hazard rate of imposing border restrictions, suggesting some sensitivity to the economic costs of these measures. There are also regional effects in all the models in Table 2, and in Model 6, the proportion of regional neighbours imposing restrictions is negatively associated with the timing of restrictions.

Interestingly, the substantive impact of nationalism on the timing of imposing border restrictions is fairly comparable to that of the perceived risk of COVID-19 spread as measured by distance from Wuhan, China (using results from Model 4). As noted above, a more nationalist regime is 2.3 times more likely than a less nationalist regime to impose restrictions on any given day during the early phase of COVID-19. States that are close to China were 2.8 times more likely to do so compared to those further away.

‘Defense in Depth’ or Security Theatre?

Finally, if the security theatre logic is operating and border restrictions are seen as a way of creating a feeling of security by associating the disease with foreigners and deflecting accountability, then nationalist regimes should be less likely to adopt border restrictions as a part of a comprehensive approach to outbreak response. A comprehensive response to provide actual security from COVID-19 would include domestic measures. To examine this expectation, I look at the timing in days of the imposition of each state’s first domestic stay-at-home order (when most people were recommended or required to stay at home) relative to each state’s first border restriction. Only a handful of states imposed domestic measures before international border restrictions. But, there is marked variation in how quickly states imposed domestic measures after imposing border restrictions. Because only two states imposed stay-at-home orders in the early phase before February 29, this analysis uses data through 2022 (when most states had already imposed their first stay-at-home order if they were going to and when the Oxford COVID-19 Government Response Tracker stopped collecting data).

Consistent with hypothesis 3 and again accounting for the role of risk of or vulnerability to COVID-19 spread, I find that states with more nationalist regimes impose domestic measures more slowly relative to international border restrictions. States with a more nationalist regime are about 40 per cent less likely to impose domestic stay-at-home orders on any given day after imposing border restrictions than those with less nationalist regimes (hazard ratio of 0.61 [0.61 times as likely], p < 0.05) (simulation based on Model 8, from Table 3).

Table 3. Cox proportional hazard models explaining the timing of states’ first stay-at-home order relative to states’ first international border restriction (presented as hazard rates)

***p < 0.001; **p < 0.01; *p < 0.05; ·p < 0.1. Note that standard errors, in parentheses, are of coefficient values, not of the hazard rates (exponentiated coefficients).

In summary, even accounting for perceived risk of or vulnerability to COVID-19 spread, nationalist regimes were more likely to impose international border restrictions in the early phase and to do so more quickly, and less likely to adopt such measures as a part of a comprehensive outbreak response strategy. These findings support the argument that governments use international border restrictions as global health security theatre and point to a key reason why many states acted at odds with WHO guidance in the early phase of COVID-19.

Robustness Checks

I explore several alternative explanations and other robustness checks (all results are fully reported in the appendix). First, I argue that a key reason why nationalist regimes are attracted to international border restrictions as security theatre is because such measures associate the disease with foreigners and deflect domestic accountability. But, do nationalist regimes come to power as a part of a broader process that includes a state’s overall approach to the border? In other words, is the relationship actually shaped by policy legacies?

As noted above, Kenwick and Simmons (Reference Kenwick and Simmons2020) suggest that states’ ‘border orientation’ impacts the stringency of border controls during COVID-19 because prior investments in border security lead governments to draw on those same strategies during a pandemic and suggest a relationship between border orientation and overall restrictiveness of border measures during the first six months of the pandemic. Does a state’s prior investment in border security explain its first border restrictions during COVID-19?

I include states’ border orientation scores as a control and find that it is negatively associated with states’ first border restriction (p = 0.101), contrary to what one would expect if states were doing what they always do at the border. Nationalist regimes remain positively associated with imposing border restrictions. While overall border orientation may be positively associated with some aspects of states’ border policies during COVID-19, this article’s findings show the salience of nationalism and global health security theatre for states’ initial border restrictions. I explore the possible role of policy legacies in two other ways. I control for whether states adopted border restrictions during H1N1 (2009) or Ebola (2014) and there is no association with states’ first COVID-19 border restrictions.Footnote ⁸ I also control for states’ ‘welcoming score’ measured via how many passports each state accepts visa-free, with a visa on arrival, or with an electronic travel authorization (Passport Index 2023). There is no association with COVID-19 border restrictions and the association between border restrictions and nationalist regimes remains.

Second, I include alternative measures of several variables in the analysis. As alternative measures of risk of or vulnerability to COVID-19 spread, I include a measure of capacity to minimize the spread of disease at points of entry (POE) (World Health Organization 2023). Note that these scores are self-reported by states and therefore subject to potential bias. I also include the level of health expenditures as a percentage of GDP and the Human Development Index (HDI) (World Bank 2022). Alternative measures of domestic political and economic pressures include trade dependence on China, overall trade exposure, and an interactive effect between outbreak preparedness and democracy, a main finding in past work on this topic. Findings are consistent. I also examine whether, instead of looking at the timing of each state’s first stay-at-home order, results are consistent when looking at the timing in days of the imposition of each state’s first internal movement restriction (such as between cities or sub-national regions) relative to each state’s first border restriction. Results are consistent. One other implication of the argument is that the role of nationalism is not conditional on a state’s political system. As such, I also include an interaction term between democracy and nationalist regime and, as expected, do not find a conditional relationship.

Third, variance inflation factors for several predictors in the full regression models suggest that multicollinearity could influence the results (for example, the regional dummies are correlated with distance from Wuhan and government effectiveness is correlated with GHSI score and GDP per capita). After removing some of these predictors from the model, the VIF for all predictors falls below 4 and the findings remain consistent.

Fourth, I address the possibility that the results are driven by certain characteristics of the data. I use multiple imputation to account for missingness and extend the survival analysis beyond the early phase, through March 2020 when all states had imposed their first border restriction and after the WHO changed its recommendations. The findings remain consistent.

An Illustration: A Case Study of the United States

The US during COVID-19 provides a useful illustration of the argument. On January 31, 2020, one day after the WHO declared the PHEIC and recommended against international travel restrictions, the US announced a restriction on entry of most foreign nationals who had been in China in the past two weeks to go into effect on February 2, 2020. A look at the Trump administration and its overall response to the pandemic illustrates the global health security theatre logic at work. I show below that the Trump administration had reason to see international travel restrictions as a particularly useful tool for security theatre because they offered a visible policy action that aligned with the administration’s underlying ideology, achieved other desired policy goals, and resonated with the administration’s key constituents. As the argument also expects, the administration not only adopted the entry restriction against travellers from China within the early phase of COVID-19, but also did so more quickly than others, and the measure was not part of a comprehensive outbreak response strategy.

The US was ranked as the most prepared country to respond to a global health emergency the year before the pandemic by the Global Health Security Index, had the 13^th largest GDP per capita in the world, and though it had high air passenger volume, was located thousands of miles away from Wuhan, China where COVID-19 was first reported. Thus, it had little reason to feel especially vulnerable to COVID-19 spread in early 2020 (though these early assessments of US capacity would later prove to have been overly confident). Along these lines, US President Trump noted on January 22, 2020, that ‘we have it totally under control’ (Owermohle Reference Owermohle2020).

Furthermore, US public health officials and agencies were not in favour of border restrictions in the days prior to the Trump administration’s announcement of the first entry restriction targeting foreign travellers from China (though some public health officials would later state support for the measure). For instance, in a briefing to senators on January 24, 2020, Anthony Fauci, Director of the National Institute of Allergy and Infectious Diseases (and later a member of the White House Coronavirus Task Force), told reporters in reference to a travel restriction on China that ‘it’s not something that I think we’re even considering’ (Sullivan and Weixel Reference Sullivan and Weixel2020). Similarly, when Deputy National Security Advisor (and later also a member of the Coronavirus Task Force) Matthew Pottinger pushed to limit travel from China in late January, he ‘clashed with CDC officials …[who] held the traditional public health view that border closures interfere with the movement of medical personnel and goods’ (Berg et al. Reference Berg, Callahan and Rotella2020). That public health officials did not support the policy suggests that the January 31 order was not based on an expert recommendation that it would protect public health.

The Centers for Disease Control and Prevention (CDC), other parts of the Department of Health and Human Services (HHS), and the President-appointed White House Coronavirus Task Force advised the President. But, decisions about border management lie primarily with the Executive. Accounts of the early days of the COVID-19 response make clear that this was especially true for the Trump administration (see, for example, Abutaleb and Paletta Reference Abutaleb and Paletta2021; Berg et al. Reference Berg, Callahan and Rotella2020). In early 2020, the US had no reason to feel particularly vulnerable to COVID-19 and public health agencies were not strongly advocating for border restrictions; but, the country had in the Trump administration a government characterized by a nationalist ideology.

According to the V-Dem data, the US ranked in the highest quartile in late 2019 for the extent to which the current government promotes a nationalist ideology. Evidence of exclusionary nationalism in the Trump administration often manifested in policy and rhetoric targeting foreigners and migrants, even prior to COVID-19. Trump’s political slogans ‘Make America Great Again’ and ‘America First’ are cases in point. Other examples include Executive Order 13769, ‘Protecting the Nation From Foreign Terrorist Entry Into the United States’ issued in 2017 restricting visas for individuals from seven countries, five of which are predominantly Muslim, and accusing Democrats of wanting migrants to ‘infest our country’ (Liptak and Klein Reference Liptak and Klein2018). Additional examples abound (Finley and Esposito Reference Finley and Esposito2020). President Trump also had a history of associating disease outbreaks with foreigners and calling for travel restrictions not recommended by the WHO. During the 2014 Ebola outbreak (before Trump was running for president), he criticized the Obama administration for not restricting travel from countries affected by Ebola, stating the following on Twitter: ‘The U.S. cannot allow EBOLA infected people back’, ‘STOP THE FLIGHTS!’, and ‘NO VISAS FROM EBOLA STRICKEN COUNTRIES’ (Werner et al. Reference Werner, Abutaleb, Bernstein and Sun2020).

The January 31 order restricting entry into the US therefore aligned with the administration’s broader ideology, as well as its policy agenda, which was often anti-immigrant and focused on ‘border security’. As such, the Trump administration also had reason to think that international travel restrictions would resonate with key constituents. Trump’s supporters tend to share anti-immigrant views (Doherty Reference Doherty2016; Scott Reference Scott2019). Indeed, racist resentment and anti-immigrant sentiments were key determinants of a Trump vote in 2016 (Hooghe and Dassonneville Reference Hooghe and Dassonneville2018). The expectation that this view would persist during COVID-19 was borne out. Gadarian et al. (Reference Gadarian, Goodman and Pepinsky2023) find that, in the US, racial resentment predicted support for international travel bans during COVID-19 and that this relationship was strongest amongst Republicans. At the outset of COVID-19, then, the US had a government characterized by exclusionary nationalism that was likely to see border restrictions as useful for security theatre because such measures reflected the administration’s ideology and policy priorities and also appealed to constituents.

The government’s rhetoric and policies during COVID-19 reflect the security theatre logic. During the pandemic, President Trump behaved like other nationalist leaders, promoting an exclusionary nationalist agenda by downplaying the threat of the virus early on, ‘appealing to exceptionalism by arguing that their country would be uniquely protected … promoting as-yet unproven preventative measures and treatments, and blaming outgroups including other countries and transnational actors’ (Givens and Mistur Reference Givens and Mistur2021, 217). Though President Trump downplayed the virus early on, he also made a point of rhetorically associating it with China. When he claimed that ‘we have it totally under control’ on January 22, he continued that, ‘it’s one person coming in from China, and we have it under control. It’s going to be just fine’ (Owermohle Reference Owermohle2020). Over time, the administration increasingly associated the virus with foreigners, China in particular. On February 2, days after announcing the entry restriction, Trump noted that ‘We pretty much shut it down coming in from China’ (Keith Reference Keith2020). In March 2020, after the entry restriction was imposed, Trump started referring to the virus as the ‘kung-flu’ and ‘Chinese virus’, with other administration officials defending the use of the term (Itkowitz Reference Itkowitz2020; Rogers, Jakes, and Swanson Reference Rogers, Jakes and Swanson2020). Illustratively, a photo of a daily press briefing in March shows a statement to be read by President Trump with the word ‘Corona’ crossed out and replaced with ‘Chinese’ (CNN 2020).

The timing of international border restrictions imposed by the US and the way in which they were imposed further supports the security theatre logic. The US announced its first entry restriction on January 31, in the early phase of COVID-19 before the WHO changed its recommendations on February 29^th. Furthermore, while the US was by no means the first to impose an entry restriction – 55 states did so earlier than the US – the January 31 announcement came before 126 states imposed their first restriction.

As the argument also expects, this border restriction was not a part of a comprehensive outbreak response plan. For one thing, the entry restriction was not implemented in a way that would maximize its public health utility. Though the entry restriction on travel from China was imposed relatively quickly compared to other states, it was not imposed quickly enough to provide public health protection. By the end of January, COVID-19 had already spread beyond China, as evidenced by the fact that ‘the large epidemic that unfolded in New York was linked to travellers from Europe’ that were not included in the first restriction (Bollyky and Nuzzo Reference Bollyky and Nuzzo2020). Travel from Europe was not restricted until mid-March when the outbreak in New York was already underway. Furthermore, the entry restriction on travel from China (as well as subsequent entry restrictions on Europe and other countries) was ‘leaky’ in that it exempted large categories of travellers, including US citizens and other ‘essential’ travel. These exemptions further reduced the public health utility of such restrictions (Bollyky and Nuzzo Reference Bollyky and Nuzzo2020).

In addition, border restrictions were not implemented as a part of a strategy that included domestic measures. For example, the US did not deploy a comprehensive testing strategy until June 24, 2020, while the median date for deployment across all states was April 16, 2020 (a comprehensive testing strategy is defined as either testing anyone with symptoms or on request) (Hale et al. Reference Hale, Angrist, Goldszmidt, Kira, Petherick, Phillips, Webster, Cameron-Blake, Hallas, Majumdar and Tatlow2021). While missteps at the CDC contributed to delays in the testing rollout in the US, the Trump administration made misleading statements about the availability of tests and did not relax regulations preventing private laboratories from developing their own tests until the end of February 2020 when the virus was already widespread in the US, and actively downplayed the threat of COVID-19 to the US public through at least mid-March 2020 (Keith Reference Keith2020; Boburg et al. Reference Boburg, O’Harrow, Satija and Goldstein2020). Similarly, at no time within the first year of the pandemic were most people in the US subject to a recommendation or requirement to ‘stay at home’. In other words, the way in which early border restrictions in the US were implemented suggests they were not viewed as one component of an overall strategy offering multiple layers of protection; the US was not practising ‘defense in depth’.

Importantly, this failure to launch a comprehensive response was not completely due to a lack of capacity or expert knowledge about what a better response should entail. As Nuzzo et al. (Reference Nuzzo, Person, Cameron, Taylor, King, Aspinall and Becker2024) note, ‘The Department of Health and Human Services and the Department of Homeland Security had developed pandemic plans, such as the National Pandemic Influenza Plan, the Biological Incident Annex to the Response and Recovery Federal Interagency Operational Plan, the Pandemic Crisis Action Plan, and others, before 2020’. Many of these plans outlined the need for a multi-layered, more ‘defense in depth’ type of response. Lewis (Reference Lewis2021) also points to a number of individuals within – or with key access to – the government in the early days of COVID-19 who were aware of the need for, and pushing for, a more comprehensive response.

The US failure to deploy its potential capacity during COVID-19 is the result of multiple factors, and a full analysis of its weak response is outside the scope of this article (see Covid Crisis Group 2023). Still, none of these shortcomings in implementation would have prevented the Trump Administration from even calling for a comprehensive response strategy on paper in the early phase of COVID-19. Instead, imposing a border restriction in isolation and touting its ability to ‘shut it down coming in from China’ further supports the contention that this measure was imposed as security theatre.

While this article focuses on states’ first border restrictions, subsequent border restrictions imposed by the US underscore the security theatre logic. The Trump administration’s use of Title 42 to expel asylum seekers and refugees at the southern border and restrictions on land travel from Mexico and Canada illustrate how international border restrictions invoked in the name of COVID-19 had little support from public health agencies and helped the administration pursue its other desired policy goals (see Garrett and Sementelli Reference Garrett and Sementelli2022). In this case, the administration’s longstanding promise to supporters to ‘build a wall’ at the southern border.

As reported by Berg for ProPublica, in mid-March 2020 Trump’s senior advisor Stephen Miller, a ‘driving force’ behind Trump’s 2017 ‘Muslim ban’ mentioned above, pushed for the administration to use CDC powers to close the southern border. When Martin Cetron, CDC’s director of the Division of Global Migration and Quarantine, received language for the proposed order, he refused to sign it. One of his deputies noted that she ‘was not a fan of trying to make the case that Canada and Mexico represent a big risk on the land border based on what we “believe” is occurring vs. what we know about the # of cases (which are far fewer than the # of cases in the US now due to community spread)’. Cetron told a colleague, ‘I will not be a part of this … it’s just morally wrong to use a public authority that has never, ever, ever been used this way. It’s to keep Hispanics out of the country. And it’s wrong’. CDC’s Trump-appointed Director, Robert Redfield, signed it instead (Berg et al. Reference Berg, Callahan and Rotella2020). The US case illustrates how the security theatre benefits of border restrictions can lead governments to adopt suboptimal outbreak response policies.

Interestingly, the US case points to a troubling dynamic whereby measures based on security theatre could trigger path-dependent processes. The approach to border restrictions did not fundamentally change when Biden became president (Worsnop et al. Reference Worsnop, Truong, Dhar, Grépin and Lee2024; see also, Tufekci Reference Tufekci2021). The Biden administration’s underlying ideology does not rest on exclusionary nationalism so COVID-19 border policy might have looked different had Biden been president at the outset. But, Trump’s early decisions about the border and pandemic response overall may have activated political incentives that narrowed the policy options going forward, making it likely that Biden would adopt a similar approach. Given that the domestic politics surrounding the pandemic response persisted after Biden entered office, the security theatre approach to border restrictions may have seemed costly to reverse as the Biden administration focused on what were seen as more important pandemic policy priorities (Maxmen and Subbaraman Reference Maxmen and Subbaraman2021; see also Worsnop and Marion Reference Worsnop, Marion, Kaarbo and Thies2024). As scholars of path dependence note, the sequence of events matters and early decisions can constrain future choices (Pierson Reference Pierson2000; Levi Reference Levi, Lichbach and Zuckerman1997). That Trump was president at the outset of the pandemic rather than Biden may have been significant from the perspective of COVID-19 border policy.

Path dependence is particularly likely to operate during a pandemic, not only because of political incentives but also because early decisions and interventions have a significant impact on the trajectory of a pandemic (see, for example, Binny et al. Reference Binny, Baker, Hendy, James, Lustig, Plank, Ridings and Steyn2021; Chinazzi et al. Reference Chinazzi, Davis, Ajelli, Gioannini, Litvinova, Merler, Pastore y Piontti, Mu, Rossi and Sun2020; Hatchett, Mecher, and Lipsitch Reference Hatchett, Mecher and Lipsitch2007). In the case of the US, that a comprehensive response was not launched early made it more difficult to bring the pandemic under control in the country later, meaning that on top of the political disincentives to changing border policy, even a significant change in the approach to the border under Biden might not have made a large difference from a public health perspective. The possibility that border measures enacted for security theatre could trigger path-dependent processes deserves further study in the US context and elsewhere.