In May 2025, the member states of the World Health Organization (WHO), the World Health Assembly, announced that it had adopted the WHO Pandemic Agreement, a multilateral treaty governing future pandemic responses (WHO 2025).The first of its kind, this ambitious endeavor creates a legally binding system of pandemic prevention and response by establishing a global network of information, knowledge, manpower, and resource sharing (Butchard and Balogun Reference Butchard and Balogun2024). The WHO features as a key coordinator for pandemic preparation and response, technical expertise, capacity development, and financial support.Footnote 1 Although in early discussions states suggested mechanisms of coercive enforcement, the agreement does not add any coercive mechanisms for enforcing its new standards (Wenham, Eccleston-Turner, and Voss Reference Wenham, Eccleston-Turner and Voss2022). Article 19 of the agreement leaves it to the Conference of the Parties, the signatories to the treaty, to “consider and approve the establishment of a mechanism to facilitate and strengthen effective implementation” of the agreement (WHO 2025, Article 19). These mechanisms are expressly to be “facilitative in nature” and shall make “non-binding recommendations” (Article 19). The Conference of the Parties also is endowed with the authority to engage in periodic review of the implementation of the agreement every five years (Article 18(4)(a)).
To be sure, the adoption of this agreement is a significant moment in the history of the WHO and an important example of global cooperation. However, the inclusion only of facilitative and non-binding mechanisms for encouraging compliance with the agreement raises again the question of the WHO’s role in coordinating pandemic policy making. If the WHO will take on a greater role in shaping policy responses to future pandemics, what mechanisms can it use to do so most effectively?
Using data collected at the start of the COVID-19 pandemic in early 2020, we show that the WHO played a leading role in coordinating policy making during the last global health crisis (Shen & Vega Reference Shen and Vega2025). The organization’s pandemic declaration on March 11, 2020 called for all countries to take action against the outbreak (WHO 2020a). Within five days of that declaration, 31 countries had closed their schools and 46 had canceled public events, measures recommended by the WHO to control the spread of the new disease.
We argue that the WHO’s prominent role in the diffusion of policies against COVID-19 worked through a leadership mechanism. In this article, we develop the concept of diffusion by leadershipFootnote 2 and expand its applicability to international organizations. Leadership differs from other diffusion mechanisms because it involves the purposeful promotion of a policy and a relational aspect with followers who decide to adopt that policy based on the leader’s legitimacy. The effects of leadership are particularly strong during a crisis, when governments search for quick guidance on how to protect their populations. In the case of measures against COVID-19, there was no time to evaluate the outcomes of policies by learning about the effects of their implementation abroad (Baybeck, Berry, and Siegel Reference Baybeck, Berry and Siegel2011; Gilardi, Füglister, and Luyet Reference Gilardi, Füglister and Luyet2009; Weyland Reference Weyland2005). Nor was there time for normative emulation that achieves compliance through the naming and shaming tactics typical of international organizations (Finnemore Reference Finnemore1996; Finnemore and Sikkink Reference Finnemore and Sikkink1998; Greenhill Reference Greenhill2010). Our research reveals that policies to contain the coronavirus diffused because of the WHO’s leading role and because of imitation between countries. The explanation for the WHO’s influence on policy makers in such a fast diffusion process derives from its position as a global leader in public health, and its impact was enhanced by the uncertainty of the situation. The organization promoted the adoption of its favored policies against COVID-19 within weeks because governments saw it as a leader in the pandemic. Rather than applying sanctions or reshaping norms, the WHO provided guidance during a crisis. In the middle of the storm, countries sought a lighthouse to illuminate their path—and found the WHO.
Our research leverages the different recommendations the WHO gave regarding five policies. Using event-history analysis, we identify the impact of the WHO’s pandemic declaration on the diffusion of these recommendations. Our results show that the WHO had a positive effect on the rate of adoption of policies it promoted, revealing the impact of its call for action and its guidance on how to deal with the new virus. Governments saw the WHO’s recommendations as coming from a trustworthy and legitimate source to guide their policy-making decisions in a moment of despair.
However, the WHO’s declaration of a pandemic had no impact on the adoption rate for policies discouraged by the organization, showcasing a limitation of the leadership mechanism in trying to prevent governmental action in a particular direction during a crisis. The dire crisis placed policy makers in a cornered situation with a pressure to act. Following recommendations not to act would have been very difficult at the onset of a global pandemic, and governments that willingly followed the WHO’s guidance on what policies to adopt decided to ignore its recommendations about which ones not to implement.
The imitation mechanism helps explain why countries did not follow the WHO’s requests that countries not restrict international and domestic travel. Like leadership, this mechanism was particularly strong in the diffusion of policies against COVID-19 and played an important role in the diffusion of the five policies we analyzed. We do not claim that the WHO was the sole or primary cause of diffusion but instead show that the organization was able to affect policy diffusion by leading governments to adopt three policies. We argue that diffusion studies should pay more attention to international organizations acting as catalysts of diffusion, even if they face significant challenges with enforcement or compliance. Especially in moments of crisis and uncertainty, their legitimacy may be enough to have an impact on diffusion through the leadership mechanism (Dobbin, Simmons, and Garrett Reference Dobbin, Simmons and Garrett2007, 456).
Leadership as the WHO’s Diffusion Mechanism in the Onset of a Crisis
Policies against COVID-19 spread across hundreds of countries in a matter of days in early 2020 (Rausis and Hoffmeyer-Zlotnik Reference Rausis and Hoffmeyer-Zlotnik2021). The fastest policy to be adopted was school closures: 157 countries decided to close schools within 56 days of the first country’s adoption, 140 of these adopted the policy within one month after first adoption, and 13 countries closed their schools before registering the first case in their country. Even the slowest policy to diffuse, workplace closures, was implemented by 104 countries within 80 days of the initial adoption. These measures spread so quickly that 41 countries adopted at least one policy before registering their first case of COVID-19; on average, they did so more than two weeks before reporting a case of the virus. The speed of diffusion, counted in days instead of years, is much faster than cases discussed in the literature of international policy diffusion—including some of the fastest ones, like the diffusion of conditional cash transfers, which was called a “tidal wave” (Sugiyama Reference Sugiyama2011) and a “surge” (Vega Reference Vega2024) because of its speed. Our contribution lies in explaining the fast diffusion of policies against COVID-19 by showing that imitation of other countries was not the only mechanism spreading these policies. The WHO’s leadership also played a significant role.
Leadership as a Diffusion Mechanism
Policy diffusion studies aim to explain why different political units adopt the same policy. In other words, what causes one policy to attract the attention of multiple governments to the point of implementing it? The political units may be countries, states or provinces, municipalities, or any governmental unit. Multiple studies theorize that internal characteristics of these units determine their chances of adopting a model from abroad (Berry and Berry Reference Berry, Berry, Weible and Sabatier2018, 265–72). Features like the severity of the problem that a policy intends to solve (Stream Reference Stream1999), state capacity (Aidt and Jensen Reference Aidt and Jensen2009), and the presence of domestic advocacy coalitions (Jenkins-Smith et al. Reference Jenkins-Smith, Nohrstedt, Weible, Ingold, Weible and Sabatier2018) all make it more likely that a government decides to emulate others in policy making. Despite the relevance of domestic features (including in the diffusion of policies against COVID-19, as shown later in the analysis), diffusion processes imply ways through which policy models are disseminated and promoted across political units.
Scholars theorize external determinants for mechanisms of policy diffusion (Berry and Berry Reference Berry, Berry, Weible and Sabatier2018, 256–60; Mooney Reference Mooney2020). The most common mechanisms in the literature are learning, imitation, competition, coercion, and normative emulation. We argue that leadership can be an important additional mechanism that acts as a catalyst for the diffusion of policies, especially in a period of crisis like the global spread of COVID-19. Furthermore, we state that international organizations can activate the leadership mechanism by encouraging countries to adopt some policies.
The conceptual definition of leadership as a diffusion mechanism begins with an understanding of political leadership using the well-established differentiation between power and leadership (Burns Reference Burns1978).Footnote 3 Our understanding builds on the notion of leadership as a behavior to influence the decisions of followers (Elgie Reference Elgie2015, 26–30; Wiener Reference Wiener1995). This conceptualization of leadership is relational, depending on a two-way connection between leader and follower. It is also an active view of leadership because the leader’s influence does not emanate as an exemplary model but instead derives from purposeful actions intended to guide others. Finally, a concept of leadership decoupled from power must be based on legitimacy, which reiterates the two-way relationship between an active and purposive leader and trusting followers.Footnote 4
As a diffusion mechanism, leadership also operates as a relationship between a leader and a group of followers. The leader could be a government, an international organization, or a transnational NGO that purposefully promotes a policy. The followers are governments that see the leader as legitimate in that issue area and decide to adopt the policy based on the leader’s recommendations. An and colleagues’ Reference An, Butz, Cha and Mitchell2023 study of the adoption of municipal climate policies identifies that some local governments have a larger impact in the diffusion process, implying that other cities see them as legitimate sources of good policy ideas. However, they still do not present the concept of leadership as requiring a particular behavior on the part of the leader.
In contrast, our behavioral conceptualization requires action on the part of the leader to promote a policy. In that sense, it shows parallels with Dobbin, Simmons, and Garrett’s (Reference Dobbin, Simmons and Garrett2007, 456) view of leadership in diffusion, which frames diffusion leaders as powerful countries that use their position to promote policies actively. That formulation identifies leadership as a purposeful type of behavior aimed at influencing others. However, the limitation to powerful countries included in their definition ignores the conceptual distinction between power and leadership, which centers on the use of legitimacy more than force and resources to influence followers. Most similar to our conception is the understanding presented by Solorio (Reference Solorio2021), who discusses Mexico’s entrepreneurial leadership in climate policy negotiations (despite the country’s weak record in implementing those policies).
The concept of leadership as distinct from power allows for the application of the concept to the role of international organizations in policy diffusion. Even without the traditional forms of hard power attributed to countries, these entities act as “transfer agents” that may facilitate this diffusion (Dolowitz and Marsh Reference Dolowitz and Marsh1996; Stone Reference Stone2004). Like catalysts, they accelerate the diffusion of policies by actively promoting them to multiple governments. As defined earlier, leadership is relational and requires followers to respond positively. International organizations’ leading behavior will only resonate with policy makers who respect the organization’s authoritative knowledge on that subject matter and accept its policy recommendations as desirable.Footnote 5 In other words, leadership from international organizations rests on countries’ views about their legitimacy. Lacking coercive powers, only organizations seen as legitimate can generate voluntary compliance of countries (Lenz and Söderbaum Reference Lenz and Söderbaum2023, 902). Government officials and bureaucrats will freely decide to follow an organization’s recommendations because they recognize its capabilities to identify and promote the most suitable policies. This legitimation rests on the organization’s recognized technical knowledge and information about an issue, as is the case with the WHO’s expertise on infectious diseases.
The Uniqueness of Leadership as a Diffusion Mechanism
In developing the concept of leadership as a mechanism of diffusion, it is crucial to clearly differentiate our argument from what is already established in the diffusion literature. First, our behavioral definition of leadership differs from a loosely defined use of the term “leaders” as referring simply to first adopters in a diffusion process. In that context, the term is opposed to “laggards,” meaning late adopters (e.g., Mallinson Reference Mallinson2021; Volden, Ting, and Carpenter Reference Volden, Ting and Carpenter2008). This terminology merely establishes that early adoption of a policy may influence other adoptions; it describes the government’s position in the curve, without explaining the mechanism(s) at play in the diffusion process. Therefore, we reject the use of that terminology in favor of leadership as a concept referring to the purposeful act of guiding and influencing governments to adopt a policy.
The mechanism of leadership, we argue, is distinct from other mechanisms identified in the literature. It differs from learning, competition, and imitation because of the centrality of the relationship within which both leader and follower must act. In those three mechanisms, the initial government does not have to do anything other than implement the policy. Other governments decide whether to adopt the same policy based on what they can gather about it from abroad, without any direct promotion of the policy. In other words, learning, competition, and imitation cannot describe the catalyst role of international organizations as transfer agents, because the concepts do not involve active promotion of the policy.
In learning, the government decides whether to adopt a policy based on the outcomes observed from the policy elsewhere (Meseguer Reference Meseguer2004). Competition refers to a government adopting a policy implemented by its peers after observing that it provides a competitive advantage to them (Baybeck, Berry, and Siegel Reference Baybeck, Berry and Siegel2011). In these two mechanisms, the diffusion process takes multiple years because of the need for potential adopters to observe the policy come to fruition abroad before they have the information needed to make a rational evaluation of its effects (Weyland Reference Weyland2005). In one example of learning, the diffusion of hospital financing reforms took 18 years to spread across OECD countries (Gilardi, Füglister, and Luyet Reference Gilardi, Füglister and Luyet2009). An example of the competition mechanism comes from the dispute among postcommunist countries related to attracting foreign investors, which led to the diffusion of flat tax programs in a process that took 21 years (Appel and Orenstein Reference Appel and Orenstein2013).
Imitation occurs when countries adopt a policy quickly based on limited information about other countries’ adoptions without these countries or other entities necessarily promoting the policy (Weyland Reference Weyland2005). Diffusion based on the imitation mechanism can be faster because there is no need to wait for the actual results of the policy. The diffusion of pension privatizations among Latin American countries exemplifies those rushed decisions (Weyland Reference Weyland2006). Our results show that imitation’s fast impact on diffusion was at play during the diffusion of policies against COVID-19 (see also Mistur, Givens, and Matisoff Reference Mistur, Givens and Matisoff2023), but the WHO’s leadership also played a role as a conceptually distinct mechanism centered on the relation between the organization and its country members.
The mechanisms of coercion and normative emulation may seem closer to leadership because they include a relational dynamic between an entity promoting the policy and potential adopters.Footnote 6 Moreover, both concepts apply to international organizations just as does leadership. However, they operate through different relational logics that allow us to distinguish leadership as a concept. The coercion mechanism describes diffusion that happens when powerful countries or international financial institutions force governments to adopt a policy (Berry and Berry Reference Berry, Berry, Weible and Sabatier2018, 259; Graham, Shipan, and Volden Reference Graham, Shipan and Volden2013, 692; Hearson and Tucker Reference Hearson and Tucker2023; Mooney Reference Mooney2020, 30). The stereotypical case is when powerful international financial institutions promote policies in developing countries using sanctions and economic incentives. Elkins, Guzman, and Simmons (Reference Elkins, Guzman and Simmons2006, 840) identify an effect of the IMF in the diffusion of bilateral investment treaties. There is no claim for legitimacy nor voluntary search for guidance on the part of policy makers: coercion is precisely the use of power that our conceptualization contrasts with leadership. Coercive diffusion curves can be as quick as several years, depending on the deployment of resources to push the policy onto weaker governments. Henisz, Zelner, and Guillén (Reference Henisz, Zelner and Guillén2005) found that countries depending on credit from the World Bank and the IMF were more likely to adopt market-oriented reforms in a diffusion process that reached more than 60 countries between the 1980s and 1990s.
Normative emulation (Mooney Reference Mooney2020, 25–29) is also dependent on the action of those willing to promote the policy. This type of diffusion, theorized by constructivist scholars (Dobbin, Simmons, and Garrett Reference Dobbin, Simmons and Garrett2007, 450–54; Finnemore Reference Finnemore1996; Finnemore and Sikkink Reference Finnemore and Sikkink1998), involves the reshaping and channeling of new ideas to change the consensus around an issue and the policy goals related to it (Greenhill Reference Greenhill2010; Hall Reference Hall1993). Once the new values are widely accepted, naming and shaming strategies are used to mobilize governments around the new global values and norms and further incentivize policy adoptions. Normative emulation is an essential mechanism for international organizations. Nevertheless, the relationship between promoter and potential adopters in this mechanism is not one of leader and follower. It depends on governments’ willingness to present themselves as valuable and respectable members of the global order, be it from a positive expectation of benefiting from it or from the negative expectation that they may be shamed and excluded otherwise. This is different from leadership, in which governments look for the leader’s guidance based on trust that it provides sound recommendations for adopters’ immediate benefit. The respect and voluntary decision to follow a leader in policy diffusion rests on the legitimacy associated with technical expertise and a specialized body of knowledge in the issue area. In turn, the need to reestablish the global perception about a policy slows down diffusion processes depending on normative emulation. Finnemore and Sikkink’s (Reference Finnemore and Sikkink1998, 896) analysis shows that it took female suffrage organizations more than 80 years to reach a tipping point. More recently, Velasco (Reference Velasco2020) reveals the impact of international organizations in reshaping norms to promote LGBT policies, but this diffusion process resulted in fewer than 30 adoptions in 25 years. Similarly, it took more than 20 years for country adoptions of electoral observers (Hyde Reference Hyde2011, 8).
In the context of healthcare policy diffusion, cases with a coercion mechanism are extremely rare.Footnote 7 Like most international organizations, the WHO cannot impose conditionalities and lacks financial powers to coerce governments (Gostin, Sridhar, and Hougendobler Reference Gostin, Sridhar and Hougendobler2015; Weyland Reference Weyland2006, 171). The WHO’s most frequent mechanism in catalyzing diffusion is normative emulation, with the long-term construction of global norms associated to values linked to the expansion of public health. The WHO played that role in the diffusion of mental health policies, using membership and peer pressure as means to encourage governments to join the bandwagon (Shen Reference Shen2014). Similarly, the WHO’s promotion of universal healthcare in the declaration of Alma-Ata helped place the topic on the political agenda, creating a normative paradigm conducive to the model’s diffusion (Linos Reference Linos2013, 8; Weyland Reference Weyland2006, 17). However, the normative emulation mechanism cannot explain the diffusion of policies against COVID-19 in early 2020 in a matter of days or weeks: this rapidity does not align with the slow policy processes associated with reshaping global norms, which take years or decades. The WHO’s leadership role, legitimized by the organization’s knowledge and expertise, was able to catalyze such a fast diffusion process.
In addition, leadership is a mechanism that is particularly suitable for periods of increased uncertainty. Together with imitation, which governments use to react quickly and without complete information in the face of a crisis, leadership diffusion gains more relevance when governments do not know what to do and look for guidance from organizations that they see as legitimate. When their previous policies and understandings seem inadequate to deal with a sudden and insurmountable international challenge, policy makers look abroad for ideas. In that process, we see both imitation and leadership. For the first mechanism, governments look at what other countries are doing regarding that same problem, which leads to policy diffusion by imitation. Like ships in a storm, countries follow each other in search of a safe harbor. They also seek the guidance of specialized agents whom they trust to know more about the problem and potential solutions to it, resulting in diffusion by leadership. In the face of the increasing number of deaths, governments sought solutions they could apply immediately to prevent the spread of COVID-19. The WHO recommended some of these solutions. Although countries certainly followed each other in the storm, as indicated by our results regarding the imitation mechanism of diffusion, the WHO had the important role of a lighthouse, recognized by the ships’ captains as a secure source of guidance to safety.
WHO’s Legitimacy and Leadership
The legitimacy of an international organization depends on its technical expertise and its ability to make that knowledge available in valuable ways for others. It also requires engagement in negotiations to provide solutions for collective action problems faced by governments. The WHO has been historically successful in these areas. The Global Polio Eradication Initiative achieved great success on a global scale and was enabled by the WHO’s provision of organizational and technical support. The WHO not only coordinated international efforts from multiple governments but also allocated thousands of staff members to more than 70 countries to help devise and implement eradication policies (Aylward and Tangermann Reference Aylward and Tangermann2011). In a more circumscribed case, the diffusion of integrated Community Case Management of Childhood Illness policy in sub-Saharan Africa was possible because “trusted partners” acted as transfer agents, using their legitimacy to support the learning and normative mechanisms (Bennett et al. Reference Bennett, Dalglish, Juma and Rodríguez2015). In our terms, the leadership mechanism operated alongside others to accelerate the model’s adoptions.
The WHO has institutional tools that regulate its relations with countries to facilitate coordination during disease outbreaks. The 2005 International Health Regulations (IHR) updated and expanded its authority deriving from the preexisting International Sanitary Regulations framework (WHO 2016). The 2005 system introduced a stricter set of standards for reporting on the spread of disease; expanded the scope of the regulations to any medical condition, not only the six initial communicable diseases; established binding duties on state parties intended to strengthen international and national healthcare capacity; and placed the WHO in a more central role in the response to the spread of diseases (Gostin and Katz Reference Gostin and Katz2016; Meier Reference Meier2022).
The regulations have a tool that establishes the WHO as a key coordinating actor during public health emergencies: the Public Health Emergency of International Concern (PHEIC). A PHEIC declaration obligates the WHO Director-General to provide temporary recommendations for countries regarding policy responses to the health emergency (WHO 2016, Article 15). These countries’ governments may adopt additional health measures that comply with national and international duties and achieve greater levels of health protection. However, they are prohibited from adopting restrictions on international travel or intrusive policies where “reasonably available alternatives” exist (Article 43.1). Between 2005 and COVID-19, the WHO Director-General had issued a PHEIC declaration four times, with successful mobilizations of collective responses, as in the cases of H1N1 and Ebola (Gostin and Katz Reference Gostin and Katz2016; Katz and Sorrell Reference Katz, Sorrell, Rushton and Youde2015).Footnote 8 However, no response matched the speed and global coverage of the diffusion of policies against COVID-19 in 2020, save perhaps the response to H1N1.
These successes did not prevent criticism and challenges of the PHEIC instrument. Some experts argued that the Director-General’s decision to trigger a PHEIC declaration was delayed or influenced by political considerations (Gostin and Katz Reference Gostin and Katz2016). Conversely, countries often adopted restrictions on international trade and travel in violation of Article 43 of the IHR (Forman and Habibi Reference Forman, Habibi and Flood2024; Meier Reference Meier2022). Countries also continued to face challenges of capacity, despite the IHR’s intention to proactively build state capacities for responding to public health emergencies (Meier Reference Meier2022).
Although the reforms of the past two decades have established a binding legal framework for responding to global health emergencies through an expanded role for the WHO as a central coordinating actor, it has faced significant obstacles in exercising that role. The efforts to develop a Pandemic Preparedness Treaty show that the WHO and member states are addressing these deficiencies by giving the WHO stronger mechanisms for encouraging coordination and compliance. In 2022, the IHR also saw a round of amendments to address perceived deficiencies revealed during the COVID pandemic.
This article’s key question is whether the WHO’s legitimacy in early 2020 allowed it to exercise its leadership role in the diffusion of policies against COVID-19. Despite the limitations of IHR regulations and criticisms of the PHEIC instrument, the WHO was among the most trusted international organizations in the world at the time. Work from the LegGov program comparing six international organizations revealed that the WHO consistently had the highest legitimacy scores among elites and citizens in all countries analyzed: 83% of the elite respondents gave the WHO a positive score, with around one-third giving it the highest score. The confidence in the WHO is steadfast among political and bureaucratic elites (Dellmuth et al. Reference Dellmuth, Scholte, Tallberg and Verhaegen2022), the two groups that led the adoption of policies against COVID-19 in early 2020. Schlipphak, Meiners, and Kiratli (Reference Schlipphak, Meiners and Kiratli2022) also provided evidence of the WHO’s strong legitimacy among citizens. Therefore, the weaknesses of the WHO’s institutional measures did not preclude it from acting as a policy catalyst at the onset of the COVID-19 pandemic, because it was seen globally as a legitimate coordinator of policy making in a public health crisis.
The WHO against COVID-19: Alerting When and How to Act
The WHO played a dual role in the diffusion of policies at the onset of COVID-19. First, it monitored the development and spread of the disease to be able to alert countries when they should act against the new disease. The pandemic declaration on March 11, 2020 was the most critical alert message the WHO sent to governments. The second role of the organization was to provide guidance regarding the policies countries should adopt to keep the virus under control. The leadership mechanism implies that countries would follow these policy recommendations when the organization called them to act.
WHO’s efforts to monitor the new disease started hours after China’s government notified it of cases of atypical pneumonia in the last days of December 2019. The organization quickly activated its Incident Management Support Team to evaluate the situation. In the following weeks, the WHO published reports and communications informing the world about the disease’s existence, the first case outside China, and the risk of an international outbreak.Footnote 9 This monitoring informed governments how much they should worry about the new coronavirus. A vital alert came on January 30, when the WHO, informed by the IHR Emergency Committee, made a PHEIC declaration about COVID-19. Despite the seriousness of this message and the legal duty of countries to respond, few countries acted immediately to contain the coronavirus. As shown later, there was little change in policy adoption following the WHO’s PHEIC declaration.
The PHEIC framework clearly outlines the WHO’s role in global public health as a voice of technical expertise and leadership. However, because the WHO did not use the PHEIC framework effectively in past outbreaks, it is important to test whether and to what extent the WHO was able to play this leadership role during the outbreak of COVID-19. For that reason, we focus here on the WHO’s pandemic declaration, which was associated with a significant increase in countries’ policy adoptions.
Unlike the PHEIC, “Pandemic” is not a legal term of art under the current framework, nor did the pandemic declaration trigger the legal obligations found in the IHR.Footnote 10 Therefore, this declaration and its contribution to policy diffusion provide an opportunity to study the WHO’s leadership role during public health crises outside the context of binding international law and the contentious discussions surrounding the PHEIC. The pandemic declaration focuses our discussion on the WHO’s leadership role.
The statement delivered in a media briefing on March 11, 2020, in which the WHO first declared COVID-19 a pandemic, was a direct call for worldwide action by the Director-General. The organization had already stated that governments should be ready to take action based on their exposure to the virus, but the pandemic declaration signaled that the whole world was at risk. WHO Director-General Tedros Ghebreyesus emphasized the rapid global spread of COVID-19 cases and called on countries to adopt policies to contain the virus immediately. Two quotes from him exemplify this urgency: “We cannot say this loudly enough, or clearly enough, or often enough: all countries can still change the course of this pandemic,” and “I remind all countries that we are calling on you to activate and scale up your emergency response mechanisms” (WHO 2020a).
Because of this evident change in the WHO’s discourse—issuing a much bolder call for action from countries’ governments—we consider March 11, 2020 to be the relevant starting point for the organization’s effect on policy adoption. We theorize that WHO’s leadership on the diffusion of measures against COVID-19 should result in more countries implementing those measures immediately after that declaration because governments relied on the organization to monitor the virus’s worldwide spread. In other words, the WHO activated the leadership mechanism to promote the diffusion of policies in the context of a global public health crisis.
In addition to telling countries when to act, the WHO’s recommendations on how to act against the new virus promoted the diffusion of several specific policies. Lacking time to observe the results of policies taken by other countries, the best that individual countries could do was imitate other governments or follow a leader. In that environment of uncertainty, the WHO published clear recommendations for governments, providing explicit guidance about what they should do. In Pamuk’s (Reference Pamuk2022) terms, the WHO maximized the usefulness of its recommendations. Even if the organization later received criticism for its response to COVID-19 (see Hernández Reference Hernández2020; Independent Panel for Pandemic Preparedness and Response 2021), it was one of the few legitimate sources of guidance on which countries could rely at the onset of the pandemic.
What did the WHO tell countries to do? This research leverages differences in the WHO’s recommended policies to estimate its impact on the diffusion of measures against the disease. Of the five policies analyzed here, the organization only recommended three: school closures, workplace closures, and cancelation of public events.Footnote 11 It formalized these recommendations in a document on February 28.Footnote 12 The “Report of the WHO-China Joint Mission on Coronavirus Disease 2019 (COVID-19)” praised China’s implementation of lockdowns in affected regions and suggested that countries with cases or outbreaks “conduct multi-sector scenario planning and simulations for the deployment of even more stringent measures to interrupt transmission chains as needed (e.g., the suspension of large-scale gatherings and the closure of schools and workplaces)” (WHO 2020b, 22). Similar recommendations for these measures were also included in a document on March 7 (WHO 2020c, 2), issued just four days before the WHO declared that COVID-19 was a pandemic. That proximity in time corroborates our methodological strategy to operationalize the organization’s impact on diffusion by focusing on the March 11 declaration of the pandemic.
Notably, the WHO also actively discouraged countries from restricting travelers, especially on international trips. As early as January 10, a WHO publication had the subtitle, “International Traffic: No Restrictions Recommended” (WHO 2020d). Multiple documents repeated this guideline. In early February, a booklet titled “Strategic Preparedness and Response Plan” argued that travel restrictions were only helpful in specific circumstances and stated that “restricting the movement of people and goods during public health emergencies may be ineffective and may interrupt vital aid and technical support” (WHO 2020e). Even as the virus spread globally, the organization advised countries not to restrict travelers, as in this guidance from February 29: “Several countries that denied entry of travelers or who have suspended the flights to and from China or other affected countries, are now reporting cases of COVID-19” (WHO 2020f). This policy advice persisted throughout the period of our dataset.Footnote 13
By comparing the WHO’s impact on the diffusion of each of the five policies, this research can compare its impact on the three policies it recommended—school closure, workplace closure, and the cancellation of public events—with the effect it had on the diffusion of policies it discouraged: domestic and international travel restrictions. The theoretical expectation is that the organization’s leadership at the onset of the pandemic convinced countries to adopt recommended policies when it declared that COVID-19 was a pandemic, while having a negative effect on the adoption of other policies.
We acknowledge that the WHO’s performance during the COVID-19 pandemic generated strong criticism. Some experts criticized the slow response (Eccleston-Turner Reference Eccleston-Turner2020). The WHO lacked enforcement powers for its health data reporting requirement, so it could not impose compliance when the Chinese state declined to report data on COVID-19 or stymied nonstate actors’ reporting (Meier Reference Meier2022). Most notably, US President Donald Trump accused the WHO of being politically motivated to support China, questioning its ability to provide unbiased recommendations. However, these accusations only occurred in the end of March, after the main period of this article’s focus. An analysis of discourse from multiple countries’ governments shows that even Trump was sympathetic to the WHO until mid-March when the diffusion processes analyzed here occurred. In addition, the same study points out that leaders from China, the United Kingdom, Germany, Japan, Australia, and France had largely positive views of the organization during the period of interest for this study (Yang Reference Yang2021). Specialists also supported the WHO’s behavior at the time (Habibi et al. Reference Habibi, Burci, de Campos, Chirwa, Cinà, Dagron, Eccleston-Turner, Forman, Gostin and Meier2020). In short, the potential harm to the WHO’s legitimacy caused by COVID-19 happened after the diffusion of policies promoted by the organization and therefore could not have influenced its ability to promote those policies through its leadership.
If the WHO had a positive effect on diffusion as a leader by declaring a pandemic, we should expect countries to adopt policies to close schools and workplaces and cancel public events. Our results confirm that hypothesis. One could also expect the organization’s leadership to reduce the diffusion of domestic and international travel restrictions, given the WHO’s recommendations for countries not to adopt those policies. However, our results reveal that the crisis and the signal of the WHO’s pandemic declaration encouraged governments to act against the new disease strongly, to the point of countering the effects of recommendations not to adopt some policies. In the discussion section, we propose that the magnitude of prospective losses caused by the pandemic made advice not to act a very hard sell. Governments were willing to take risks in attempts to protect their populations and imitated other countries’ example when it came to the policies not recommended by the WHO.
Data and Methods
The rapid diffusion seen in 2020 has multiple simultaneous causes and mechanisms. Our methodological strategy, with five event-history models, controls for the multiple causes and isolates the effect of the WHO’s recommendations. We do not claim that WHO was the only or even the most important cause of diffusion. Our research shows that, despite its lack of strong enforcement, the WHO influenced governments’ decisions regarding when to adopt measures against COVID-19 and what measures to adopt, contributing to the fast diffusion of policies it promoted.
Event-history models are well suited to study diffusion (Gilardi, Füglister, and Luyet Reference Gilardi, Füglister and Luyet2009; Moehlecke Reference Moehlecke2020). They estimate variables’ effects on the time it takes for each unit to experience an event—in our case, adoption of each of the five policies. All models in this article follow the Cox regression because we do not establish parameters about the relationship between covariates and time (Box-Steffensmeier and Jones Reference Box-Steffensmeier and Jones2004, 47–48).Footnote 14 We apply an event-history model for each of the five policies analyzed: school closures, workplace closures, cancellation of public events, restrictions of domestic travel, and international travel bans.
Dependent Variables
The dependent variables in our models are the time it takes for each country to adopt one of the five policies. However, the new coronavirus reached some regions of the world earlier than others. For example, the disease was not a concern outside East and Southeast Asia until late January. Therefore, we apply a different starting point for the dependent variable based on the occurrence of COVID-19 in its region.Footnote 15 For each country, the starting point (t = 0) is the day when the number of cases in its region reaches 10 (table 1). This starting point ties countries’ varying perceptions of the disease over time to the immediate threat of the disease close to them. It includes policies that were adopted before the country registered its first case.
Starting Dates per Region
Using region-based starting dates also avoids length bias for regions in which COVID was late to appear. This type of bias can artificially reduce hazard rate estimates (Wolkewitz et al. Reference Wolkewitz, Allignol, Schumacher and Beyersmann2010, 206–7). A single starting date also could bias the estimates for variables characteristic of early adopters, as found in East and Southeast Asia countries where COVID was first identified.Footnote 16
This project uses data from numerous sources that we compiled. For our data on policy adoptions, we rely on the Oxford COVID-19 Government Response Tracker dataset (Hale et al. Reference Hale, Petherick, Phillips and Webster2020). For school closures, we rely on data from UNESCO (2020).Footnote 17 Only the governmental determination to implement a policy (e.g., closing schools or banning international travel) is counted to obtain a uniform measure of national adoption. We exclude subnational adoptions because they indicate decision making by subnational actors and subnational variation in policy adoption, although we do control for level of subnational policy making in our models. With this simplification, the operationalization of each policy is a daily measurement of adoption/nonadoption at the national level, identifying the day of adoption.
We select the five policies based on their similarities to each other and ease of implementation. First, none of the policies required technical expertise or particularly strong state capacity. Second, although they affect social life, these policies did not affect the entire population, so their implementation required reduced policing power. Third, governments were able to implement all of them almost immediately, and they were straightforward policies that did not allow much adaptation or variation. These characteristics reduce concerns about potential alternative explanations. Governments’ decisions to implement these measures determined their adoption—not state capacity, technical bureaucracy, strength of the executive, established infrastructure, or even time for adaptation. Despite that rationale, we analyze three additional policies from the Oxford dataset that required larger state capacity for implementation: closure of public transport, stay-at-home requirements, and additional protections for the elderly. Results for these three policies corroborate our findings about the WHO’s leadership on policy diffusion.Footnote 18
Table 2 describes the pattern of policy adoptions within the time frame of our dataset. Although states had begun to implement policy responses to COVID-19 before the March 11 pandemic declaration, there was a greater number of policy adoptions in school closures, workplace closures, and public event cancellations during the five days after the WHO declaration than in the nearly two-and-a-half months preceding it. Notably, the number of travel restrictions also increased after the pandemic declaration, though not with the same dramatic impact. Domestic restrictions grew steadily after the declaration, whereas many countries adopted international travel restrictions even before the WHO declared COVID-19 a pandemic.
Timing of Policy Adoption (Average Rate of Adoption/Day in Parentheses)
Explanatory Variables and Empirical Expectations
The most important variable in our models refers to the WHO’s pandemic declaration to estimate its impact on countries’ decisions to adopt the policies. It is a binary variable identifying the five days including and following the declaration on March 11. As discussed earlier, this was the organization’s most meaningful sign that all countries should respond immediately to the virus. A fast policy adoption within five days after that declaration indicates that the WHO’s call for action led a government to implement the measure. This variable therefore identifies whether the hazard rate of adopting the policy increased right after the organization prompted countries to adopt policies against COVID-19.
We hypothesize different effects from the WHO’s pandemic declaration, depending on whether the organization promoted a policy or not. We expect a positive effect on the increase in the number of adoptions for policies promoted by the WHO (school closures, workplace closures, and cancellation of public events). Conversely, we expect a negative effect on the number of adoptions for policies not promoted by the WHO (restrictions to domestic travel and international travel bans). Our findings corroborate the first hypothesis, revealing the leadership mechanism at play in the diffusion of those three policies. However, the models show no significant effect in the case of policies discouraged by the WHO. We explain this surprising finding based on the fact that a pandemic declaration still signals that countries should act, potentially triggering more imitation and the adoption of well-known responses like travel restrictions despite the WHO’s advice.
Beyond the WHO’s pandemic declaration, the models include a series of other variables to control for alternative potential causes of diffusion and confounders.Footnote 19 Because imitation is an essential driver of diffusion, especially in a crisis like the COVID-19 pandemic, we include two variables to account for the number of countries that have already adopted the policy each day. The effects of imitation measured by these variables were already identified in previous research (Mistur, Givens, and Matisoff Reference Mistur, Givens and Matisoff2023), so we expect them to have a positive coefficient for all policies. Our two variables differentiate between global and regional imitation. The first variable considers the number of adoptions in the world, whereas the second one counts the number of adoptions in each country’s region to identify whether countries tended to imitate their neighbors more than countries far away (Brinks and Coppedge Reference Brinks and Coppedge2006; Mooney Reference Mooney2001).
Two other variables measure the global scale of the crisis: the number of cases of COVID-19 and the number of deaths caused by the disease each day (JHU CSSE 2020; WHO 2020g). Countries received these figures daily, which gave them some indication of the COVID-19 threat. Previous research about subnational policy making against the new coronavirus in the United States had mixed results about the importance of those numbers for decision makers (Adolph et al. Reference Adolph, Amano, Bang-Jensen, Fullman, Magistro, Reinke, Castellano, Erickson and Wilkerson2022; Hartney and Finger Reference Hartney and Finger2022).
We include variables related to the United States and China to estimate the effects of global powers on diffusion. Following the same logic of the pandemic declaration variable, two variables account for five days after the US and Chinese governments adopted the policy. Countries influenced by those powers should follow them and adopt the same measures. Two other variables identify countries with strong exposure to US or Chinese influence, drawing on Levitsky and Way’s (Reference Levitsky and Way2010, 40–42) argument that economic linkages alter policy making. A free-trade agreement with the United States was a proxy for exposure to the United States, and membership in the Belt and Road Initiative pointed to ties with China.
We also build our models to account for internal determinants of policy making by considering the threat of COVID-19 within each country and each country’s institutional capacity to respond to that threat. The models include four country-specific variables. One is the number of COVID-19 cases in the country (JHU CSSE 2020; WHO 2020g). Countries with more cases should decide faster to incur the costs of policies against the new coronavirus. The second variable is the number of airplane passengers per year, as measured by the World Bank. The third, also from the World Bank, is the proportion of the population older than 65. These three variables identify countries at higher risk of suffering significant losses because of the novel coronavirus. Finally, each country’s healthcare capacity is likely a strong determinant of its policy response to the pandemic, given that countries should assess their risk in a pandemic based on their ability to respond to outbreaks among the population. We include a modified version of the Global Health Security Index, which measures countries’ preparedness for epidemiological threats (Cameron, Nuzzo, and Bell Reference Cameron, Nuzzo and Bell2019). Our modified variable avoids correlation with the dependent variable and emphasizes characteristics directly linked to healthcare capacity.Footnote 20
An additional set of variables controls for ideological and populist tendencies in governments to evaluate claims that right-wing populist leaders responded more slowly to COVID-19.Footnote 21 Two are adapted from Norris’s (Reference Norris2020) dataset, and two others come from V-Dem (Coppedge et al. Reference Coppedge, Gerring and Knutsen2020). We also have variables measuring different aspects of democracy to control for bottom-up pressures that could foster or hinder the adoption of policies. One identifies countries that had elections scheduled for the remainder of 2020, which could affect leaders’ behavior. Considering previous findings that countries led by women acted more efficiently against COVID-19 (Aldrich and Lotito Reference Aldrich and Lotito2020; Funk Reference Funk2020), we include data on the proportion of women in parliament. Two other variables from V-Dem measure the power of regional and local governments to account for within-country variations in subnational policy adoption. Finally, four other variables provide standard controls: the gross domestic product, Human Development Index, population, and land area.
The findings from these statistical models are reinforced by a series of alternative models and robustness checks presented later in the article and in the appendices. These checks show that our findings are robust to model specification and model fit and that our model satisfies the proportional hazards assumption of the Cox regression. A random forest survival analysis shows that the timing of the WHO’s pandemic declaration, along with variables measuring imitation, accounts for much of the variation in the timing of policy adoption for school closures.Footnote 22
Results
The sequence of adoptions of each policy followed the typical S-shaped curve of diffusion processes, beginning with a small number of adoptions, then entering a steep part of the curve with multiple adoptions in a short period, and reaching a plateau after most countries had already implemented the policy. Figure 1 shows the diffusion curves for all five policies.
Adoption Curves in Relation to WHO’s Pandemic Declaration (Vertical Line)
Notes: Diffusion curves based on the number of policy adoptions per day. The x-axes count the number of days passed since the beginning of the timeframe for analysis (January 1, 2020). The vertical line marks March 11, 2020, the date that the WHO declared COVID-19 to be a pandemic.
The graphs in figure 1 show that the steepness of the curves is related more or less closely to the WHO’s pandemic declaration on March 11. Our statistical analysis confirms the significance of that relation for the three policies promoted by the WHO: schools and workplaces closure and cancellation of public events. The probability that countries closed schools and workplaces and canceled public events within five days of the WHO’s pandemic declaration is significantly higher than for the rest of the period. These are the three policies promoted by the WHO. This timing shows that countries became more concerned about COVID-19 once the WHO declared it was a global crisis and followed its lead regarding policy recommendations to contain the virus.
The pandemic declaration variable lacks statistical significance for the two policies discouraged by the organization: domestic travel restrictions and international travel bans. The coefficient for each of these policies is not statistically different from zero. This finding challenges our expectation of a negative coefficient but aligns with our observation in figure 1 that domestic travel restrictions and international travel bans increased around the same time as the pandemic declaration, which went against WHO’s guidance. All estimates in this section report coefficients, not hazard rates, and therefore, estimates over zero reflect an increase in the likelihood of policy adoption (figure 2).
Coefficient Plots for the WHO’s Effects on All Policies
Countries’ compliance with three of WHO’s recommended policies is impressive, considering its lack of enforcement power. As theorized, the context of crisis heightened the effectiveness of the organization’s leadership mechanism based on its legitimacy. The declaration of the pandemic increased governments’ perception that something had to be done immediately against COVID-19, and they responded by following the WHO’s communications about the most suitable policies to reduce the spread of the disease.
In addition to the impact of the WHO’s leadership mechanism on the diffusion of policies against COVID-19 in the first weeks of 2020, our models also identified other causes for these policies’ adoptions. Table 3 presents the results for all policies.
Results from Cox Regression for All Five Policies
Notes: Control variables included: GDP, HDI, land area, population, local governance, and regional governance.
Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
Alternative Explanations for Policy Adoption
The models show that usual diffusion patterns following imitation and responses to risk also took place. The WHO did not replace the usual mechanisms of diffusion but instead was a transfer agent that supplied a menu of potential policy responses and acted as a catalyst promoting these policies. Variables measuring imitation—the number of policy adoptions globally and regionally—showed that the actions of other governments had a significant impact on policy adoption. Governments actively pursued ideas of how to protect themselves by looking at other countries’ examples, even those outside their region. Although the WHO found followers willing to adhere to its guidance, it was not the only source governments used to inspire their response.
The threat of COVID-19 itself affected the timing of policy adoptions: the global number of deaths caused by the disease sped up adoption, whereas the global number of cases had a significant impact in slowing down diffusion. These estimates imply that most governments were not concerned by the initial rapid increase in the number of cases in China and neighboring countries. It was only when the number of deaths increased rapidly elsewhere in mid-February and when the WHO issued its pandemic declaration in March that policy makers realized they were in a global crisis and had to act. Overall, countries looked outward when deciding how to prepare for COVID-19. Their policy-making decisions followed the number of deaths, imitated other countries, and responded to the WHO’s recommendations.
State capacity did not affect decisions to adopt the policies: our measure of national health capacity was not statistically significant in any of the models. This is an odd finding if we expect policy makers to evaluate their policies based on healthcare readiness or responsiveness. However, if we examine this pattern through the lens of crisis-oriented policy making and diffusion by leadership, this finding is consistent with our argument that governments reacted to the risk seen abroad by following the WHO’s guidance and imitating other countries more than evaluating their local risk and bureaucratic capacities.
Other Variables
Two other explanatory variables open avenues for further research. Economic ties with China had a negative effect on policy adoption. Membership in the Belt and Road Initiative (BRI) correlated with a delayed adoption of all five policies, but the slower response among these countries did not signify an unwillingness to adopt the policies. Although BRI countries took longer on average to respond to the pandemic, they eventually adopted all five policies in similar or greater proportions as others.Footnote 23 The effects of these “linkages to the East”Footnote 24 are relevant at the onset of COVID-19 because the virus began in China. Governments had to balance the risk of a closer relationship with the source of the disease and concerns about harming their ties with a country that provides aid and investment through the BRI. Also surprising was the speed with which these ties were established. China’s growing ties are already transforming how countries make domestic policy, even though BRI was less than a decade old at the time of the pandemic declaration.Footnote 25
Interestingly, countries with more women in parliament were quicker in adopting all five policies. This result corroborates the idea that countries governed by female leaders responded more efficiently to COVID-19 and is in line with studies on the role of women in responding to COVID-19 (Aldrich and Lotito Reference Aldrich and Lotito2020; Funk Reference Funk2020).
Robustness Checks
A series of robustness checks and alternative model specifications reinforce our results. We summarize these robustness checks here in brief and in detail in the appendices.
First, we used a random forest survival analysis to run iterations of different models, drawing combinations from all explanatory variables. This analysis generates variable importance scores that indicate which variables account for a higher degree of variation in the timing of policy adoption for school closures. Our test takes the mean of variable importance scores from 20 random forest survival analyses focusing on school closures. The WHO’s pandemic declaration variable is among the variables with the highest mean variable importance score, reflecting its importance as a signaling mechanism alongside the variables measuring the spread of similar policies. This finding corroborates our central argument, that the WHO promoted the diffusion of school closures through its leadership. For brevity, figure 3 shows only the 15 variables with the largest variable importance scores.Footnote 26 In addition to the WHO’s pandemic declaration variable, variables measuring global and regional diffusion through imitation consistently account for the most variation in the rate of policy adoption. This points to the importance of imitation as a mechanism for diffusion across the whole period of the analysis.
Mean Variable Importance Scores from 20 Random Forest Survival Models
Second, we tested the proportional hazards assumption on which the Cox model relies. That assumption states that the rate of change is constant with respect to time. Failure to account for this assumption can bias estimates. We tested this assumption using Schoenfeld residuals and applying Grambsch-Therneau tests for independence between time and Schoenfeld residuals to show that the proportional hazards assumption holds (Park and Hendry Reference Park and Hendry2015).Footnote 27 We conducted this test for the WHO Declaration variable in each policy model, cross-referenced with transformation of time. Presented in Table 4, the results show that residuals for the WHO Declaration variable have no relationship with the basic time variable, t, at a p<0.05 confidence level for all five policies. Therefore, the proportional hazards assumption holds for our Cox regressions. We also examine the relationship between covariates and time with models that discard the proportional hazards assumption using the Weibull, Gompertz, and exponential regressions. None of these models alter our findings (see appendix F).
Grambsch-Therneau Tests of Proportional Hazards Assumption, Using Schoenfeld Residuals for the WHO Pandemic Declaration Variable from Five Policy Models
Notes: Reporting χ2 values*** p<0.01, ** p<0.05, * p<0.1.
Third, we applied an alternative model specification, adding interaction terms between the WHO declaration variable and variables measuring the number of global policy adoptions and of adoptions in a country’s geographical region. Our findings using interaction terms between these measures of leadership and imitation confirm the importance of both variables in explaining the speed of adoption.Footnote 28
Fourth, we examined robustness to outliers. We find no pattern of error when checking deviance residuals, suggesting that outliers are not driving our results (Box-Steffensmeier and Jones Reference Box-Steffensmeier and Jones2004). Using deviance residuals, we examined whether a cluster of outliers exists around particular observations or durations. We plotted these residuals against geographical regions. Figure 4 reflects no discernible pattern of residuals over time or region for our school closures policy model.
Deviance Residuals for School Closures, Plotted over Time and Region
Note: See appendix H for deviance residuals for our four remaining policy models reflecting similar patterns.
Finally, we used Cox-Snell residuals and likelihood-ratio tests to diagnose issues of model fit (Box-Steffensmeier and Jones Reference Box-Steffensmeier and Jones2004).Footnote 29 For each of our policy models, we used the likelihood ratio tests to identify a different fitted model only with covariates that show a statistically significant effect on model fit in the likelihood ratio tests. We discarded covariates that did not contribute to model fit, generating fitted models unique to each policy. In each of five fitted models, the effect of the WHO pandemic declaration remains significant for school closures, workplace closures, and public event cancellations, corroborating our main findings.Footnote 30
Discussion
Our findings show that the WHO’s declaration led to the swift adoption of certain policies. In the early months of the pandemic, some countries such as China and Italy bore the brunt of the crisis, whereas others had not yet experienced the uncontrolled spread of the disease. Most governments had not adopted measures in response to the virus until the WHO’s pandemic declaration and its prescription of policies helped accelerate the adoption of policies. As we expected, the moment of crisis heightened the importance of leadership. Policy makers were eager for information about quick and easy ways to protect their countries, and the WHO’s recommendations came as clear guidelines they could easily apply.
Importantly, the fact that the WHO’s pandemic declaration triggered this fast policy adoption, and not the prior PHEIC instrument formally activated on January 30, shows how leadership at critical times can facilitate the adoption of solutions to crises even outside established institutions. It was not the institutionalized enforcement tool based on accepted norms but the strong and clear political message that mobilized governments. These governments followed the WHO because of its legitimacy in calling for action in the case of a pandemic and not because of the organization’s institutional rules.
However, this diffusion by leadership had limits. The WHO was successful in facilitating the fast adoption of policies based on its recommendations. Yet, it failed to suppress the adoption of policies it saw as harmful or as violating the principles of global health law, including both domestic and international travel restrictions. Rather than the expected negative relationship, our findings show no statistically significant relationship between the pandemic declaration timing and the adoption of these policies. In short, it seems like governments ignored the WHO’s recommendations not to adopt them. The diffusion of domestic travel restrictions and international travel bans, therefore, was driven by other mechanisms at play, especially imitation.
Explaining the limitations of the leadership mechanism remains a challenge. More research may be needed to understand the dynamics of multiple diffusion mechanisms pulling in different directions at once. In the case of the travel restriction policies, imitation played a greater role than leadership in facilitating policy diffusion. We present three potential explanations that open the way for further investigation.
First, the fact that travel bans are such an embedded component of states’ policy response to the spread of infectious disease might indicate that there may be a limit on the effectiveness on leadership to change states’ minds about policy choices that are already established in policy makers’ toolkit. Countries have a long history of adopting travel restrictions in the face of public health threats. This occurred even though, in previous global health crises, the WHO consistently warned against travel restrictions because of a lack of evidence about their effectiveness and their detrimental impact on the transit of healthcare providers, medicine, and equipment. In fact, the WHO’s legal framework prohibits these measures (WHO 2016, see Article 43.1). Nevertheless, governments have consistently declined to comply with these proscriptions (Forman and Habibi Reference Forman, Habibi and Flood2024; Meier Reference Meier2022). For both domestic and international travel restrictions, imitation was strongly at work. The number of policy adoptions in the world and in the country’s geographical region was not only significant but also had some of the largest coefficients among all five policies tested. After seeing other countries restricting travel, policy makers might have followed their example because they already trusted that policy, regardless of what the WHO recommended.
Second, prospect theory (Kahneman and Tversky Reference Kahneman and Tversky1979) may help us understand why decisions to adopt a policy were easier to make than those not to do so. In dire periods when there is so much to lose, governments are more likely to do whatever it takes to protect their populations. This risk acceptance in the domain of losses (Linde and Vis Reference Linde and Vis2017; Weyland Reference Weyland2002) could explain why the same countries that followed the WHO’s leadership regarding what policies to adopt were so quick in dismissing its recommendations not to implement other policies. In sum, policy makers faced with the prospect of losses caused by a disease that could kill thousands of their people would rather fail by action than by inaction, which made it easier for them to ignore the WHO’s recommendation not to adopt some policies. During crises, recommending that one stand still is a difficult sell.
Third, decisions to restrict travel were not based on a belief that these measures would work to curb the spread of COVID; policy makers knew that the WHO saw travel bans or restrictions as ineffective. Nevertheless, governments’ rationale for implementing those policies included a political calculation that doing so would generate a sense of security in the populace. In Worsnop’s (Reference Worsnop2025) terms, restricting travels was a “security theater” meant to instill safety in the population. Our results do not support her findings that nationalist governments were more prone to close borders. However, her argument that these measures were inspired by the need to portray a proactive stance in the face of the pandemic remains plausible. Government officials worried not only about how safe their people were but also about how safe they were in their own positions. Restricting travel could have been a response to popular demands for action, even for taking steps not in line with the expert and legitimate recommendations from the WHO.
In addition, imitation continued to drive policy diffusion, alongside the WHO’s diffusion by leadership. The imitation mechanism was relevant for the diffusion of all five policies that we studied. The number of countries that had previously adopted a policy largely influenced future adoptions by other governments, in line with the results of a previous study (Mistur, Givens, and Matisoff Reference Mistur, Givens and Matisoff2023). Interestingly, diffusion was not regionally bounded. Previous adoptions in the same region were relevant for school closures, domestic travel restrictions, and international travel bans but not for workplace closures and cancellation of public events, whereas worldwide adoptions were significant in all models. However, where regional adoptions were significant, they carried an effect twice as strong as that of the global number of adoptions.
Our findings related to domestic and international travel restrictions highlight the limitations of diffusion by leadership, engendered by trust in the WHO’s expertise when providing policy prescriptions based on its legitimacy. If states have a policy response in their toolkit that they have used before and other states are adopting that response quickly, then they may ignore WHO’s recommendation not to implement that policy.
This effect was particularly strong because the WHO, by declaring a pandemic, called governments to act against COVID-19 while simultaneously requesting them to refrain from taking actions regarding travel restrictions. The usual story of diffusion through imitation explains why policy makers will ignore the dissent of one actor, even an influential one like the WHO, when observing a cascade of policy adoptions. In the alternative, if states are casting about for policy solutions and are unsure about what path to follow, the influential voice of an expert organization will be effective in motivating the adoption of policies. The role of catalyst in the positive promotion of a policy was strong, showing the power of the WHO’s legitimacy in that regard.
Final Considerations
Lynch, Bernhard, and O’Neill (Reference Lynch, Bernhard and O’Neill2022) recommended that political scientists study the effects of COVID-19 using observational data on the global exposure to the disease. This article corroborates their argument that this type of research can expand the knowledge of public health specialists, with a focus on the role of political institutions in healthcare. Our article develops the concept of diffusion by leadership, showing how political scientists benefit from this type of research, even in well-studied phenomena like policy diffusion.
Our analysis identifies the WHO declaration’s substantial effect on the diffusion of policies against COVID-19. We show that the WHO was able to mobilize countries to adopt measures to contain the new virus because of its legitimacy in public health policy debates. Governments followed the organization’s call for action when it declared the disease a pandemic. They also listened to its recommendations about which policies to adopt. Countries were more likely to implement school closures, workplace closures, and cancellations of public events within five days of the pandemic declaration. In turn, the organization asked countries not to adopt domestic travel restrictions and international travel bans. Governments ignored these recommendations after the WHO defined COVID-19 as a pandemic, revealing limitations of the leadership mechanism in a context of crisis when imitation dynamics are strong.
Our research encourages a reevaluation of international organizations’ role in policy diffusion, especially in moments of crisis. In addition to exercising coercive mechanisms, such as the IMF and the World Bank do, or facilitating the slow transformation of norms to shift policy paradigms, international organizations can cause rapid diffusion waves through leadership. In a period of uncertainty, when a new disease was about to spread around the world, countries turned to the WHO before deciding when and what to do. When the world is engulfed in crisis, governments are eagerly looking for potential solutions abroad. Even with limited mechanisms to incentivize compliance, international organizations work as lighthouses guiding countries in a sea of despair and misinformation. Emphasizing the leadership mechanism can help policy makers working in international organizations prepare to assume leadership roles during crises. The role of the pandemic declaration, rather than the PHEIC, in triggering such a rapid wave of policy adoptions within days emphasizes the potential power of this leadership mechanism, which may be greater than existing institutional measures during a crisis. Further research about subnational diffusion may analyze whether domestic and transnational NGOs might play a similar catalyst role in the diffusion of policies between states, provinces, or municipalities.
Our findings also point to factors that deserve more research in the context of crises. The effect of linkages with China in slowing the policy response to the pandemic indicates the importance of studying economic connections with autocratic powers and their effects on policy. Finally, our analysis highlighted the role of women in policy making against COVID-19. Further research on policy diffusion should consider the role of gender.
Supplementary Material
The supplementary material for this article can be found at https://doi.org/10.1017/S1537592725101898.
Acknowledgments
We are grateful to Mark Hayward, Jay Kao, Carolina Moehlecke, and Kurt Weyland for thoughtful feedback about this project in its earlier stages. We are also thankful for Zach Elkins’ careful reading and constructive criticisms of an earlier version of this article. We appreciate the comments from Jack Greenberg, Daniel Slate, Magic Wade, and Sarah Wagner and from the audience in a panel at MPSA. Finally, we thank the journal editors and three anonymous reviewers for their productive engagement with our research.
Data Replication
Data replication sets are available in Harvard Dataverse at: https://doi.org/10.7910/DVN/NUYUB6
Open access
