The Sincerest Form of Flattery: Nationalist Emulation during the COVID-19 Pandemic

Abstract

As COVID-19 rapidly spread across the globe, every government in the world has been forced to enact policies to slow the spread of the virus. While leaders often claim responses are based on the best available advice from scientists and public health experts, recent policy diffusion research suggests that countries are emulating the COVID-19 policies of their neighbors instead of responding to domestic conditions. Political and geographic considerations play a role in determining which countries imitate one another, but even among countries that are politically or geographically distant, nationalist regimes seem to favor certain approaches towards the pandemic. We investigate why this is the case by examining whether countries that embrace a nationalist ideology are more likely to emulate the COVID-19 policies of similarly nationalist regimes. We demonstrate that, even after controlling for domestic circumstances and linguistic, trade, geographic, and political connections, nationalist countries are emulating each other’s responses. These results are robust and shed light not only on new mechanisms of policy diffusion but also on the growing international cooperation of nationalist regimes and leaders.

In an effort to combat SARS COVID-19, countries around the world have rapidly adopted a host of policies to slow the spread of the pandemic, many unprecedented. Yet, variation in the timing and stringency of these measures makes it difficult to discern the underlying factors driving these decisions. Social distancing policies, which try to slow or stop the spread of the virus by keeping people apart and other anti-pandemic measures could be enacted based on the best available advice from scientists and public health experts. This advice should take into account a variety of factors such as healthcare capacity, demographic vulnerabilities, and exposure to the virus from abroad. Recent research, however, has suggested that countries may be emulating the COVID-19 policies of their neighbors and peers instead of responding to domestic conditions [1]. Additionally, politics play an important role in countries’ reactions to COVID, making it unclear whether these factors are driving policy decisions.

Internationally, the politicization of COVID-19 does not seem to be a case of a traditional left-right division. There is no obvious relationship between governments’ position on a left-right political axis and their response to the pandemic. Socialist Cuba’s effective lockdown is dramatically different from the disastrous failure of its close socialist ally Venezuela [2, 3]. Relatively conservative governments in Germany, Austria, and Ireland enacted timely lockdowns, as did the Centrist French government and Social Democrats and Labor governments from Finland to Canada to New Zealand. Yet, a large number of governments lead by strongly nationalist leaders seem to have lead slower or less effective responses to COVID-19. While these leaders are mainly characterized as far right, the more salient characteristic may be a preference for nationalism over internationalism [4]. Popular media, some scholars, and public-health philanthropist Bill Gates have variously argued that nationalist tendencies have reduced the effectiveness of some countries’ pandemic responses [5, 6]. While we do not attempt to verify claims about the success of specific approaches to COVID-19, we are able to show that nationalist countries are emulating each other’s approaches.

Across the world, nationalist governments that are extremely varied in terms of their levels of development, political systems, and governance capacity, seem to fall into similar patterns in their responses to COVID-19. We argue that governments in these countries are taking cues from each other when making decisions about how to respond to COVID-19. Specifically, we hypothesize that variation in COVID-19 policies is, in part, driven by nationalist governments emulating each other. In order to test this hypothesis, we turn to the policy diffusion literature to investigate how policies spread between governance units such as countries.

This article follows the following format: First, we examine the global resurgence of nationalism in recent years and how nationalist countries and leaders communicate, emulate, and cooperate. Second, we consider nationalist responses to disasters. Third, we address how nationalist regimes and leaders have responded to the COVID-19 pandemic and possible emulation between them. Fourth, we turn our attention to policy diffusion, the theory and literature that will enable us to test our hypothesis. Fifth, we look at the data and methods to test our hypothesis. Sixth, we analyze our results. Finally, we conclude by reviewing our findings, interpreting their meaning, and pointing out directions for possible further research.

Nationalism

Globally, nationalism is on the rise. Many authoritarian and hybrid-authoritarian countries have experienced increases in nationalist sentiment and nationalist leaders have come to power in some of these countries. Yet, the more noticeable change has been in democracies. This holds true for the world’s oldest democracies, the United Kingdom and United States, as well as its biggest, India. Other democracies that appear to have experienced a recent rise in nationalism include the Philippines, Japan, South Africa, and Turkey. Even where nationalists have not come to power, they have managed to notably increase their vote share in countries including Sweden, France, Germany, Austria, and the Netherlands [7].

Following Bieber, we define nationalism as “a malleable and narrow ideology that values membership in a nation more highly than belonging to other groups. i.e. based on gender, political ideology, socio-economic group, region, seeks distinction from other nations, strives to preserve the nation, and gives preference to political representation by the nation for the nation.” [8] We also follow the distinction between the banal nationalism described by Michael Billing and the revolutionary nationalism which this new tide appears to represent. Banal nationalism works to normalize nations to citizens and foreigners alike and is evident throughout daily life including in flags, stamps, maps, and countless everyday objects and actions [9]. Revolutionary nationalism, however, is a “virulent nationalism that rejects the status quo and seeks to reassert the will of an imagined community over a political or cultural space is different from, but draws on, endemic nationalism” [8]. It is almost entirely the later which concerns us here and that we mean when we use the term nationalism.

Because nationalists prioritize membership in their own nation and seek “distinction from other nations [8]” the idea of nationalist movements in countries across the globe cooperating is, at least partially, self-contradictory. The Pan-Nationalism of the late Nineteenth and early Twentieth Centuries made sense primarily as a movement committed to decolonialization and in opposition to racialized empires [10]. Yet, some form of nationalist internationalism has been around since at least the nineteenth century and a new wave seems evident in communication and cooperation between nationalist regimes and movements today [11]. Additionally, Mikecz has argued that nationalism can be surprisingly versatile depending on how a nation defines its national identity [12].

Nationalists generally define themselves in opposition to an “other” or outgroup and part of the logic of contemporary nationalists is to define themselves as opponents of internationalism, Intergovernmental Organizations (IGOs), “…vaguely defined cosmopolitan elites, venues for cross-border collaboration, and technocratic expertise more generally” [13]. These targets, along with a “hostility to minorities and scorn for multiculturalism and pluralism,” [11] gives a growing nationalist internationalism a set of common enemies by which to define themselves. This logic has been evident in the Pandemic in many forms, but most notably in criticism of the World Health Organization (WHO) and the United States’ defunding of it [14]. Steve Bannon provides an illustrative example in both his anti-internationalist rhetoric and his recent move to Europe where he internationalized his ideology by launching a movement against the European Union [15]. The ultimate goal of this nationalist internationalism is “global cooperation among supposedly homogeneous, organically grown, closed national communities” [11].

Despite the fact that “nationalist groups around the world are building alliances and operating more and more in transnational institutions,” nationalist internationalism “remains less studied than the conventional socialist and liberal variants” [11]. This makes analyzing this emerging phenomenon of cross-border cooperation between nationalists both vitally important and difficult, especially for a large-scale quantitative approach. Additionally, the relative newness and rapid growth of nationalist internationalism mean that its influence on policy diffusion may not have been detectable even a few years earlier. Given the wealth of high-definition data it has created, the Pandemic offers a unique opportunity to test the level to which nationalist countries take cues from one another.

Disaster Nationalism and COVID-19

While previous research on policy diffusion during the pandemic found that countries in general are emulating each other’s responses, the rise of nationalist internationalism and what appears to be specifically nationalist responses to COVID-19 raises the question of whether and how nationalists are emulating each other’s responses.

Disasters are exogenous shocks that can rapidly alter the physical, human, political, social, and economic landscape of a country. They create “unsettled times, or ontologically insecure moments when social and political institutions are in flux” [16]. In some cases, this can facilitate civic engagement [17], bridging divides and prompting strangers or even enemies to go to great lengths to help each other [18]. While most natural disasters are regional, a disaster that is sufficiently widespread or severe enough to be perceived as national in scale can bolster feelings of nationalism. As former premier of China Wen Jiabao famously wrote in response to the Sichuan Earthquake: “Disasters regenerate a nation. 多难兴邦” [19]. Savvy governments may capitalize on this rally around the flag effect [20–22]. However, early evidence from the pandemic suggests that calls to nationalism may only work on those already predisposed to nationalist ideology [23].

In other cases, disasters can exacerbate existing social, political, religious, ethnic, or other divides even to the point of armed conflict [24]. This is particularly true when leaders use a disaster to demonize an outgroup, such as domestic minority groups, refugees, or immigrants [25, 26]. Nationalists also tend to blame other countries or transnational actors. Yet even when leaders do not fault outgroups for disasters, political focus tends to shift towards national problems. As a result, efforts at a more expansive foreign policy may wither as a country’s focus turns to domestic concerns. Japan may have fallen into this trap after the Fukushima disaster [27] even as its East Asian rival China avoided it following the Sichuan earthquake and rapidly expanded the scope of its foreign policy engagement [28–30].

So how does a nationalist lens help us understand countries’ responses to COVID-19? Bieber expresses concern that the pandemic will provide an opportunity for far-right nationalist governments to pursue prior preferences: ramping up authoritarianism, reducing democratic freedoms and civil liberties, promoting biases against certain groups, strengthening borders, deglobalizing, and appealing to the politics of fear [25]. Yet, these were already existing tendencies and tell us little about how nationalists will handle the pandemic itself and where such regimes look for information about and examples of policies.

It seems obvious that the COVID-19 pandemic is meaningfully distinct from most disasters we have seen over the past century. While its death toll, to date at least, is not particularly high compared to some of the largest pandemics, natural, and manmade disasters over the past century, its scope is unprecedented. The Great Chinese Famine (1958–1962) killed somewhere between 15 and 45 million people, but was restricted almost entirely to China [31]. Taken in total, the Second World War had a higher death toll and greater geographic range, but still left many parts of the world, most notably sub-Saharan Africa and Latin America, largely untouched. The two obvious points of comparison with COVID-19, especially as its death toll continues to rise, are the AIDS epidemic and the 1918 Influenza Pandemic [32]. The AIDS epidemic, while similarly spread throughout the world, was much slower moving and disproportionately impacted marginalized communities. In the early years, this made it easier to ignore the issue or blame it on outgroup scapegoats [33]. The 1918 Influenza Pandemic is more similar in speed to the spread of COVID-19, yet the world in which it occurred was so different that it is hard to draw lessons from that pandemic. Among other things, more limited and slower travel, the end of the First World War, and globe-spanning empires created a very different international political landscape from that of today [25]. Furthermore, robust data describing countries and their response to the onset of Spanish Flu are unavailable, making it difficult to examine the drivers of countries’ responses. Because the COVID-19 Pandemic is truly global in a way that few disasters have been, it is reasonable to expect that reactions may be significantly different than in more localized disasters. It might be harder or less tempting to blame an outgroup when the whole world finds itself in nearly identical circumstances. The availability of high-definition data make it possible to systematically investigate how countries responded COVID-19, allowing us to investigate for the first time the factors driving decision-making among nationalist countries in a global crisis.

Nationalist Responses and Emulation Language

Because we use an index measuring the overall strictness of anti-COVID-19 measures to operationalize nationalist emulation, the specific nature and details of nationalist leaders’ responses to the Pandemic are not addressed directly by our data analysis. However, a brief look at the responses of countries that scored above the median on our measure of nationalism (as discussed below) demonstrate certain similarities in approaches to the pandemic. Specifically, nationalist regimes often appear to have used a combination of (1) downplaying the problem, especially early on, (2) appealing to exceptionalism by arguing that their country would be uniquely protected, (3) promoting as-yet unproven preventative measures and treatments, and (4) blaming outgroups including other countries and transnational actors. President Bolsonaro of Brazil called COVID-19 “a little flu” [34] and claimed that Brazilians “are uniquely suited to weather the pandemic because they can be dunked in raw sewage and ‘don’t catch a thing’” [35]. The United Kingdom considered a herd immunity strategy, saw the virus as only a “moderate risk” [36] and refused to emulate its Northern European neighbors, allowing bars and nightclubs to stay open as the rest of Europe closed down [37]. President Duterte of the Philippines assured the media on February 3rd that “even without the vaccines it will just die a natural death” [38]. Iranian Supreme Leader Ayatollah Ali Khamenei tweeted that “COVID-19 is not such a big tragedy…” “The prayers of the pure youth and pious are very effective in repelling major tragedies” [39]. Belarusian President Alexander Lukashenko appealed to national identity when he advised Belarusians to drink vodka and visit saunas to stay healthy, while refusing to order a lockdown [40]. Hungarian Prime Minister Viktor Orban blamed foreigners and migration for the spread of coronavirus in in his country [41]. The Presidents of both the United States and Brazil focused on the same anti-malarial drug despite a lack of evidence, demonstrating nationalist internationalism by discussing a joint research effort [42].

Nationalist regimes have also tended to boast about their response to COVID-19 and, when previous overly optimistic predictions proved false, have massaged statistics to make their handling of the virus look better [34]. Turkmenistan has taken this strategy to its logical conclusion, denying the country had even a single case even while insisting its citizens wear masks to protect against “dust” [43]. Sometimes making use of these doctored statistics, nationalist leaders have tried to emphasize the success of their response, “...we are working rather smoothly and emerging from this situation with the coronavirus confidently and, with minimal losses...” Vladimir Putin told Russian state TV in June [44]. “So far we’ve done it better than nearly any other country in the world and I assess that by the end of this we will be the best in the world,” claimed nationalist Prime Minister of Israel Binyamin Netanyahu in March, only to see cases spike again a few months later [45]. The responses of nationalists seem markedly different from the cautious and measured messaging of less-nationalist leaders. The case in point was New Zealand where, despite managing perhaps the most successful Pandemic response anywhere, Prime Minister Jacinda Ardern told her citizens that “there is no widespread undetected community transmission in New Zealand… but we must remain vigilant if we are to keep it that way” [46].

There has also been evidence of nationalist leaders communicating with each other and praising each other’s responses. In March, US President Donald Trump announced a ban on travel from the 26 countries in the European Schengen Area. Yet, he made a point not only of excluding the United Kingdom (UK) from the ban, but claiming that the UK “has got very strong borders and they are doing a very good job” despite the fact that the UK was taking a relatively lax approach to the Pandemic and had more cases than many Schengen Area countries. Some speculated that this was a favor for or in admiration of nationalist Prime Minister Boris Johnson [47]. Later the two agreed on a coordinated response to the pandemic [48]. Trump is also reported to have frequently spoken with nationalist leaders including Duterte [49], Putin [50], and Erdogan [51] during the pandemic. Considering that none of these countries are particularly geographically, economically, or scientifically close, this is strong evidence of communication between similarly-minded nationalist leaders. Possible limits of international nationalist solidarity were also on display when it came to nationalist leaders’ assessment of other nationalists’ pandemic responses; Putin has publicly criticized the US handling of the pandemic [44]. This points to the inherent contradictions of nationalist internationalism which allows or encourages nationalist leaders to boost their image by criticizing outsiders, including other nationalists, even as they emulate each other.

The specifics of nationalist responses to COVID-19 deserve further study. In the subsequent sections we take one important step in this direction by assessing whether or not regular communication and stylistic similarities in nationalists’ political responses reflect deeper similarities in policies and whether or not they emulate each other in the enactment of these policies.

Policy Diffusion

Policy diffusion theory seeks to explain why and when governance units, (from counties to countries) adopt specific policies [52]. It considers internal determinants such as the problem characteristics in a country, political and institutional context, available resources and government capacity as well as external determinants in the form of coercion, competition, learning, and emulation from other governance units [53]. The combination of these factors drives the spread of policies between different governance units [54–56]. Trade flow, geographic proximity, and cultural or linguistic similarities are typically used as proxies to capture relationships between countries [57–60]. Yet, the field is still rapidly evolving. The impact of different diffusion mechanisms has been shown to change over time and recent research has demonstrated the importance of ideology to be increasing relative to that of geography [61]. But, exactly which mechanisms lead to a governance unit adopting policies is still a source of debate [62].

In the typical policy diffusion model, governance units look to each other’s policies, analyzing outcomes in order to follow successful policy experiments and adopt the optimal policies. The speed of the COVID-19 Pandemic has short-circuited this model. The rapid spread of the virus, the need to act quickly, and the lag time between the implementation of policies and statistics that reliably capture their effects mean that rather than learning from their experiences, countries are often simply emulating neighbors’ policies without information about their outcomes. Existing research has suggested that in the Pandemic, emulation primarily occurs between geographic and political neighbors (countries with similar levels of democracy) [1]. In order to test whether the same type of emulation is occurring between nationalist regimes, we follow existing policy diffusion research in conducting panel regression analysis that attempts to model the spread of policies based on how similar different countries are in terms of levels of nationalism.

Materials and Methods

We employ a dyadic event history analysis approach to analyze a panel of international COVID-19 data that we construct from a variety of secondary sources. Dyadic event history analyses are useful for explaining where diffusing policies originate [63]. Pandemic data are taken from the Oxford COVID-19 Government Response Tracker (OxCGRT) which provides daily information on international COVID-19 rates, deaths, and the dynamic government responses to the crisis. Since the rise in cases across missing observations can be consistently estimated given the observations before and after the missing data, we interpolate missing data points for the number of confirmed COVID-19 cases in each country over time. OxCGRT provides up to date indicators measuring countries’ health and economic policies as well as how they respond to the pandemic in key policy areas relevant to transmission of the virus. Eight of these policy areas (school closings, business closings, transportation shutdowns, gathering size restrictions, internal travel, international travel, event cancelations, stay at home orders, and public information campaigns) are directly linked to countries’ efforts to stem transmission and have been combined into the COVID-19 policy stringency index in the OxCGRT dataset [64]. This measure describes the level of policy stringency in each country on each day and serves as our dependent variable. As a robustness test, we additionally run our model on a more precise index focused specifically on social distancing policies which was also used in recent policy diffusion research [1]. This more precise index is important for our purposes because it excludes international border closures, which we might expect to be connected to high levels of nationalism. We develop this more specific measure through refining the original stringency index to a social distancing index by aggregating measures for the five indicators most relevant to social distancing efforts: school and business closings, transportation shutdowns, event cancelations, and internal travel restrictions.

We combine the OxCGRT data with measures of countries’ political characteristics from the Varieties of Democracy (V-Dem) dataset [65]. V-Dem is currently the standard choice for these types of comparative politics analyses. V-Dem assesses governments’ legitimization strategies based on expert opinions and provides a measure of countries’ level of nationalism along with other critical socio-political controls. The consensus opinion among expert coders of whether a country promotes nationalism as a legitimating ideology is measured on a scale of 0 to 1 in the V-Dem dataset and is described in Table 1.

Table 1.

Nationalism Descriptives

Variable	Description	n	Mean	Std. Dev.	Min/Max
Nationalism	Level of nationalism (Scale of 0 to 1)	26,325	0.57	0.31	0 1

We also control for countries’ administrative capacity using the Worldwide Governance Indicator (WGI) 2019 estimate of government effectiveness [66] and several relevant socio-economic controls. From the World Bank, we include Gross Domestic Product (GDP) per capita adjusted for international Purchasing Power Parity (PPP) and international tourism arrivals into a country [67]. From the World Health Organization (WHO), we include hospital beds per capita, healthcare spending as a percentage of GDP, and the percentage of the population over 65 years old [68].

Since we are interested in understanding how countries emulate one another in a global crisis, we develop several measures of interconnectivity to investigate diffusion across different signaling pathways. This allows us to explore how different countries’ policymakers take cues from each other and why they do so. We create international connectivity matrices from the V-Dem data as well as the Correlates of War Project (COW) and the Centre d’Etudes Prospectives et d’Informations Internationales (CEPII) to examine country interconnectivity across geographic proximity, trade, language similarity, political similarity, and ideological similarity. We focus on countries’ ideological similarity to investigate whether nationalist states emulate one another, measuring nationalism in two ways.

First, we measure the similarity between each dyad of countries in our dataset as the absolute value of the difference between their V-Dem nationalism scores. We treat this measure to ensure that we only capture similarity between nationalist peers (as opposed to similarity between more internationalist countries) by estimating similarities only among countries in the dataset with above-average nationalism scores. If we included countries with below-average scores as well, our results would indicate the impact of ideological similarity rather than nationalistic similarity. Second, we interact the nationalism scores of each country dyad to generate a paired nationalism-similarity index and use this measure to test the robustness of our results, again ensuring that our findings are not driven by emulation between more internationalist peers. While this method allows us to examine country dyads, the structure of our data prevents us from assessing the directionality of individual changes between countries. Since our dependent variable represents an index of policy stringency, it does not allow us to track the origin of specific COVID-19 policies. Even if we do not take these measures to focus only on emulation between more nationalist regimes, however, our results still appear to be robust.

Countries in our data are considered geographically adjacent if they are contiguous or separated by less than 400 miles of uninterrupted water [69]. This measure allows us to estimate the impact of regional proximity on countries’ COVID-19 responses. Economic interdependency is captured through directional trade between country dyads [70]. Political similarity is measured as the difference between two countries’ V-Dem Electoral Democracy Index. Next, we use international language similarity scores from CEPII [71] as calculated based on ethnologue classifications of language trees [72, 73]. We weight relevant COVID-19 variables by each type of international connectivity to investigate diffusion effects across each pathway. By including these dimensions of international interconnection in our model, we are able to assess the impact of countries’ nationalist ideologies while controlling for alternative pathways of influence.

Finally, we calculate the mean policy stringency and COVID-19 per capita case numbers among each countries’ neighbors for each group and lag our time-dependent variables by one day to account for a minimum period of time necessary for policy implementation. Additionally, we interact country-level time-invariant control variables with confirmed COVID-19 cases to estimate their policy impact conditional on pandemic circumstances. By interacting our country-level covariates with the countries’ confirmed cases we can assess how these variables influence a country’s policies conditional on the status of the pandemic. We merge these data into an unbalanced panel dataset covering 196 countries from January 1st to July 13th in 2020, organized using country-day as the unit of analysis. Descriptive statistics are in Table 2.

Table 2.

Descriptive Statistics

Variable	Description	n	Mean	Std. Dev.	Min/Max
Policy Stringency	COVID-19 policy stringency index	27,061	46.14	35.08	0 100
Social Distance Stringency	Social distancing policy stringency index	26,334	5.34	4.73	0 12
COVID-19 Cases	Confirmed COVID-19 cases per capita	19,521	8.44e-04	2.35e-03	0 3.52e02
Nationalism	Nationalism as a Legitimating Ideology	18,995	.0003844	.0013715	0 .02931
V-Dem	Level of democracy (Scale of 0 to 1)	18,995	4.34e-04	1.04e-03	0 1.17e02
Government Effectiveness	Level of government effectiveness (Scale of −2.5 to 2.5)	18,913	5.81e-04	2.05e-03	−4.42e-03 2.22e-02
IGO	Number of intergovernmental organizations the country is involved in	27,161	70.60	18.75	6 121
% over 65	% of total population over 65 years old	19,213	8.35e-03	1.97e-02	0 0.18
Hospital beds	Hospital beds (per 1000 people)	19,432	2.7e-03	7.06e-03	0 7.85e-02
Arrivals	International tourism, number of arrivals (millions)	19,247	1.19e04	4.60e04	0 6.68e05
GDP	Natural Log of GDP per capita translated to PPP (2020 international $)	19,181	8.64e-03	0.03	0 0.41
Geography	Weighted per capita
Neighbor Stringency	Mean social distancing policy stringency index of neighboring countries	24,558	5.55	4.58	0 12
Neighbor Cases	Mean number of confirmed COVID-19 cases in neighboring countries	23,865	5.87e-04	1.49e-03	0 2.65e-02
Economic	Weighted by imports. Millions 2014 $.
Economic Stringency	Mean social distancing policy stringency index of economic peers	26,517	1.71e03	2.04e05	0 4.51e06
Politics	Weighted by political similarity
Politics Stringency	Mean social distancing policy stringency index of political peers	26,325	87.28	91.29	0 3.38e03
Language	Weighted by lingual similarity
Language Stringency	Mean social distancing policy stringency index of language peers	17,090	4.16	5.32	0 41.01
Nationalism	Weighted by nationalist similarity
Nationalism Stringency	Mean social distancing policy stringency index of nationalist peers	13,649	38.53	36.46	0 310.34
Interacted Nationalism Stringency	Mean social distancing policy stringency index of nationalist peers using interacted nationalism measure	18,329	2.08e06	1.34e07	0 1.96e08

The structure of our data prevents us from using a directed dyad approach, creating the potential for bias in our results. Unconditional dyadic approaches can bias results, artificially strengthening the claim that emulation is taking place [74]. While the contextual evidence we present mitigates concern over our findings, future research should investigate the directionality of emulation among nationalist peers. Additionally, duration dependence and unobserved international heterogeneity over time may introduce bias to our results [75, 76]. The baseline likelihood of countries adopting more stringent COVID-19 policies may change over time as advice from world health officials or agencies change or public pressure for change shifts in response to pandemic conditions. While the short time period we observe (seven months) and the overall rapid pace of the pandemic may decrease the size, if not the likelihood, of changes to the baseline likelihood, we cannot rule out that possibility.

Fixed effects models are useful for controlling for unobserved heterogeneity among groups [77] and determining causality using panel data [78]. We use linear fixed effects regression with robust standard errors to estimate the impact of nationalist ideology on countries’ COVID-19 policy responses controlling for a diverse set of socio-political, economic, and health factors. Equation 1 describes the formal equation for Model 1 where X_it is a set of explanatory variables, α_it represents fixed effects controlling for subject-specific unobserved heterogeneity across countries, and μ_it represents the error term, varying over countries (i) and days (t). This method controls for individual-specific correlation between country-level variables in our model and allows us to draw accurate inferences [79].

$${\mathit{\text{stringency}}}_{\mathit{it}}=X_{\mathit{it}}\beta +\mathit{\text{nationalist}}{\mathit{\text{peer stringency}}}_{\mathit{it}}+{\alpha }_{\mathit{it}}+{\mu }_{\mathit{it}}$$ 1

Model 1 We examine COVID-19 policy diffusion between nationalist countries to understand how the nationalism of a country influences the choices its policymakers make. Models 1 and 2, included in Table 3, estimate the peer effects of ideologically similar, nationalist peers on policy emulation. In Model 1, we test whether nationalist countries are significantly more likely to emulate their nationalist peers, and in Model 2, we test whether this influence appears to persist after controlling for diffusion across geographic, economic, political, and lingual pathways of international interconnection.

Table 3.

Impact of Peer Effects on COVID-19 Policy

Variables	Model 1	Model 2
R-squared	0.747	0.856
Countries	118	77
Observations	16,710	11,080
Policy Stringency	Coefficient. Robust Standard Error.
COVID-19 Cases	32,211***	23,017
	(11,910)	(21,990)
Nationalism	3464*	2001
	(1834)	(1469)
Democracy	−349.4	980.1
	(4262)	(4782)
Govt Effectiveness	3411**	3488**
	(1689)	(1345)
IGO	83.56	77.81
	(61.73)	(62.71)
% Over 65	−480.8	−521.5**
	(344.7)	(253.2)
Hospital Beds	−299.2	492.3
	(554.1)	(450.8)
Arrivals	4.11e-06	−7.72e-07
	(3.00e-05)	(1.91e-05)
GDP	−3465***	−2702
	(1199)	(2168)
Neighbor COVID−19 Cases	1040	-1080
	(821.0)	(778.7)
Nationalist Peer Stringency	0.184***	0.0664***
	(0.00919)	(0.0132)
Geographic Peer Stringency		0.590***
		(0.0652)
Trade Peer Stringency		−5.26e-08
		(1.10e-07)
Political Peer Stringency		0.00278
		(0.00289)
Lingual Peer Stringency		0.0202
		(0.0303)
Constant	10.92***	4.745***
	(2.177)	(1.644)

*** p < 0.01, ** p < 0.05, *p < 0.1

Results

The results depicted in Fig. 1 (which correspond to Model 2 in Table 3, but with all independent variables standardized to improve legibility of the graph) shows the impact of key socio-political, economic, and diffusion variables on countries’ COVID-19 policy stringency. This provides a visual representation of how countries enact more stringent policies to contain the spread of the virus if their nationalist peers have done the same, even when controlling for other potential pathways of imitation.

Fig. 1.

Coefficient Estimates and 95% Confidence Intervals from Model 2

(Full statistical output is included in Table 3).

Both models in Table 3 demonstrate the influence that nationalist decision-makers have had over their ideological peers during the COVID-19 pandemic. When testing the peer-effects of nationalism individually in Model 1, we find that policymakers in countries with above-average levels of nationalism emulate the COVID-19 policies of other nationalist states. Countries with above-average nationalism are significantly more likely to implement a policy change the day after an ideologically nationalist peer changes its COVID-19 policies. Furthermore, the results from Model 2 indicate that this influence is very robust. The influence of nationalist peers persists even after controlling for other international pathways of policy emulation that have been shown to drive policy decisions during the COVID-19 pandemic.

Our results demonstrate that nationalist peer relationships significantly influence COVID-19 policy stringency even when controlling for emulation of geographic, trade, political, and lingual peers. Geographic connections remain a significant driver in our model as well, revealing support for similar influence among geographic neighbors. However, nationalist similarity appears to be a more important driver of policy than political, trade, or lingual connection. During the pandemic, nationalist countries are making policy decisions about how to respond to the global crisis by taking cues from other nationalist countries.

We conduct several robustness checks to test the validity of our claims. First, we estimate our models on a narrower policy index of social distancing policies; then, we test the influence of nationalist peer effects using an alternative measurement of nationalist similarity by interacting countries’ nationalism scores. These results are included in Models 3 through 6 in Table A of the Appendix. With the exception of Model 4, measuring the impact of our alternative measure of nationalist similarity on overall policy stringency, all of these tests uphold our initial findings. Several other models which used different configurations of control variables, although not shown, also suggest that these results are very robust.

Interestingly, while COVID-19 cases per capita is significant in Model 1, the baseline number of confirmed COVID-19 cases in a country is not significant once we control for the full set of emulation pathways in Model 2. Government effectiveness and the proportion of the population over 65 years old are significant, indicating that these variables influence policy decisions contingent on the number of confirmed cases, but countries' COVID-19 status is otherwise insignificant. Our results show that the number of confirmed cases per capita in neighboring countries is also insignificant in policymakers’ decisions (though not in all specifications). This reveals the importance of imitation and policy emulation in policymaking related to this global crisis, something previous research had already suggested [1]. Countries’ COVID-19 policies are more heavily influenced by the decisions of their geographic and ideological peers than the status of the crisis itself. Rather than responding to health statistics, countries are imitating their neighbors, while nationalist countries are mimicking their ideological peers.

The relationship of internal, country-level characteristics, as conditioned by the number of COVID-19 cases in the country, are largely insignificant in our models. These variables demonstrate policymakers’ reaction to problem conditions in their country, but the lack of broad significance reinforces the indication that decision makers are responding more to the actions of peer states than to relevant health and social characteristics at home. Aside from the peer effects of nationalist and nearby countries, we find that government effectiveness is, perhaps unsurprisingly, significantly correlated with higher COVID-19 policy stringency contingent on confirmed cases. GDP is significant and negative in Model 1, but insignificant in Model 2. This suggests that less wealthy countries feel the need to enact stricter policies in response to COVID-19 cases and may be conditional on our already controlling for government capacity. However, the percentage of a counties’ population older than 65 years old, a demographic widely considered at-risk for COVID-19, is actually negatively correlated with policy stringency in Model 2. Countries’ COVID-19 policies reacting to the presence of COVID-19 cases were significantly less stringent the higher the proportion of the population is in this at-risk category. This counterintuitive result is inconsistent with expectations and emphasizes the disproportionate influence of peer imitation over internal characteristics in this global crisis. Domestic nationalism was only weakly positively significant in Model 1, demonstrating that this is more a story of nationalists emulating their perceived peers rather than a nationalist ideology driving a specific approach to the Pandemic. Other internal characteristics are insignificant in our main model.

Discussion and Conclusions

Our findings demonstrate that nationalist countries around the world have been emulating each other’s responses to the COVID-19 pandemic. These results are robust even when controlling for emulation among similar political systems. The fact that nationalist countries are emulating each other’s policy responses to COVID-19 rather than following the latest health and scientific evidence, or even imitating other countries with similar levels of democracy, seems remarkable. Upon close consideration, there were already some signs of nationalists admiring and emulating each other’s political styles. Even before the pandemic, many leaders, for example, saw the increasing role of nationalism in the United States as a signal that they could play up their own nationalist tendencies [80].

Our analysis shows that emulation between nationalist regimes is occurring but does not provide much insight into why. There are several likely possibilities as to how this policy diffusion is happening, but more research is needed to test how important any of them are. One possible explanation for how and why nationalist regimes would emulate each other is that they think it might help them deal with the fallout from their initial downplaying of the crisis. It is possible that other nationalist regimes ramping up anti-pandemic measures provides political cover for nationalist leaders to do the same without admitting their earlier mistakes and may make leaders feel secure that they will not suffer major political consequences. Given the highly-compressed timeline, however, our preferred explanation for why nationalist leaders emulate each other is more straightforward and consistent with the previous research on this issue; the unprecedented speed and scale of the pandemic has put leaders all over the world on their heels. Their responses, therefore, often demonstrate the sometimes-unreflective emulation of countries or leaders that they see as similar to themselves [1].

Our findings do not tell us a great deal about modes of nationalist response to disasters, nor are they meant to. Instead, they show that nationalist countries are looking to each other rather than, or at least in addition to, countries with which they share geographic, trade, political, or linguistic ties. This is consistent with, but not identical to, Motadel’s nationalist internationalism, which emphasizes cooperation. The emulation evident in our model does not necessarily imply cooperation between nationalist countries, though there is evidence of communication between leaders, but it does show that they see each other as peers. Perhaps even more important than cooperation, however, we show nationalist governments are enacting similar policies based on their emulation of each other.

Our findings provide critical insight in the field of policy diffusion, both on the narrower issue of considering the impact of nationalism and on the broader view considering different policy vectors. While trade, geography, and cultural, political or linguistic similarities are likely to continue to be important mechanisms for policy diffusion, new vectors need to be explored [57–60]. Many thus far unknown vectors of policy diffusion may exist and could be relatively easily testable, especially making use of the data from V-Dem and other similar sources.

Here we have shown that a nationalist legitimating ideology is a, previously unacknowledged, vector of policy diffusion. Given the recent rise of nationalist leaders in a variety of countries and the relative newness of nationalist internationalism, this may be an almost entirely new vector of policy diffusion. It could also be that this vector was important in the past but was never tested, or that it has been important in certain eras, such as the 1930s, but not others. Evaluating which of these is true should be possible using historical data. It is also possible that nationalist diffusion will be limited to certain types of policies such as trade, refugees, or immigration. Considering the role it played in the pandemic we consider this unlikely, but again this could be a fruitful area for future research. Looking forward, it will become ever more important to ask if nationalist diffusion will continue to increase and, if so, whether it will diminish the importance of other kinds of diffusion.

What our findings mean for the future of the pandemic or nationalism generally is less clear. It seems possible that as the crisis drags on, the evolving situation and politics in each country will lead to different approaches to the pandemic. As COVID-19 becomes a known quantity, the urgency and uncertainty of the early months of the pandemic which may have left leaders with little choice but to emulate each other will wane. Nationalist leaders may plot more strategic paths, which could diverge from each other. What is likely to continue, however, is a general strengthening and deepening of ties of nationalist internationalism. Even if individual nationalist leaders, especially in democracies, lose power as a result of mishandling the pandemic, it seems likely that a general trend of nationalist leaders looking to each other for support and as examples will continue. Indeed, if nationalist leaders are willing to stick to their ideology and emulate each other even in the face of a once-in-a-century pandemic it seems almost certain that they will continue to do so in the future. It seems likely therefore that only a profound electoral defeat or other major domestic political setbacks will dampen the nationalist internationalism we have found evidence for here.

Biographies

John Wagner Givens

, Kennesaw State University, jgiven11@kennesaw.edu. John has conducted research and published on a variety of topics including the Chinese legal system, Chinese foreign policy, online politics in Asia, the Malaysian media, legal advice websites, and smart cities. John earned his BSFS from Georgetown’s School of Foreign Service, his MA in Asian Studies from UC Berkeley, and his doctorate in Politics from the University of Oxford. He previously held positions at the University of Pittsburgh, Carnegie Mellon University, the University of Louisville, and the University of the West of England. He also works as an expert in China-related legal cases.

Evan Mistur

, University of Texas at Arlington, evanmistur@gatech.edu. Evan is an Assistant Professor in the College of Architecture, Planning, and Public Affairs at the University of Texas at Arlington. He has a Ph.D. in public policy from the Georgia Institute of Technology where he specialized in energy and environmental policy. His primary research focuses on public administration and environmental management, examining how decision-making processes impact conservation in public agencies.

Appendix

Table 4.

Robustness Checks

Variables	Model	Model 4	Model 5	Model 6
	Alternative Nationalism Measure. Individual Model.	Alternative Nationalism Measure (Full Model.	Alternative Dependent Variable. Social Distancing Index.	Alternative Dependent Variable & Nationalism Measure
R-squared	0.200	0.846	0.715	0.822
Countries	89	58	118	77
Observations	12,777	8460	16,707	11,078
Policy Stringency or Social Distancing Index
COVID-19 Cases	29,178	2087	5797***	3792
	(23,781)	(30,194)	(1624)	(3502)
V-Dem	7820***	1488	536.5**	232.9
	(2650)	(2113)	(244.9)	(241.5)
Nationalism	4803***	−740.6	125.1	−111.9
	(1300)	(1049)	(115.1)	(114.7)
Govt Effectiveness	−18,537*	1519	−53.94	−636.8
	(9813)	(5312)	(546.0)	(629.3)
IGO	2165	1719	429.5**	424.8**
	(2364)	(1299)	(211.2)	(172.8)
% Over 65	251.9**	84.42	7.106	12.84
	(123.3)	(91.96)	(8.647)	(9.548)
Hospital Beds	649.1**	−518.0*	−64.13	−70.54**
	(312.6)	(274.5)	(41.11)	(34.36)
Arrivals	−857.4	669.5	−47.47	9.745
	(577.7)	(560.9)	(67.75)	(61.59)
GDP	−6.10e-05**	−6.54e-06	3.95e-06	2.97e-06
	(2.45e-05)	(2.68e-05)	(3.97e-06)	(2.36e-06)
Neighboring Cases	−4140*	−695.8	−573.2***	−377.3
	(2406)	(2877)	(159.1)	(341.6)
Nationalist Peer Stringency	8.72e-07***	−4.06e-07	0.190***	0.0796***
	(2.18e-07)	(3.12e-07)	(0.00848)	(0.0147)
Geographic Peer Stringency		0.720***		0.517***
		(0.0721)		(0.0734)
Economic Peer Stringency		−6.01e-08		−3.03e-07**
		(1.33e-07)		(1.19e-07)
Political Peer Stringency		0.00819**		0.00434
		(0.00345)		(0.00368)
Lingual Peer Stringency		0.105**		0.0168
		(0.0524)		(0.0253)
Constant	48.62***	5.183***	1.116***	0.478***
	(0.654)	(1.779)	(0.239)	(0.170)

*** p < 0.01, ** p < 0.05, *p < 0.1