Evaluating Territorial Authority over Policy Responses to the Crisis: A Comparative Study of the COVID-19 Pandemic

Abstract

This article examines the allocation of authority over policy responses to the COVID-19 pandemic in 13 advanced democracies. It introduces a new measurement scheme that allows for a systematic cross-country analysis of authority structures. The paper also tracks changes in territorial arrangements over time to investigate their responsiveness to the evolving nature of the pandemic. The analysis reveals four main patterns, including fluctuating (de)centralisation efforts that mirror changes in the urgency of the crisis, institutional incentives for coordination acting as a functional alternative to centralisation, the influence of broader territorial arrangements, and regional resistance to centralisation in politically charged contexts.

Keywords:

• policy responses
• COVID-19 pandemic
• crisis
• centralisation
• policy coordination
• territorial authority

Introduction

The outbreak of COVID-19 in early 2020 required governments to act swiftly. With vaccines still in the early stages of development, governments relied on numerous protective policies to contain the exponential growth of the virus, including the closing of borders, schools, restaurants, and cultural institutions, and banning events and gatherings. The outbreak accentuated the fragility of countries’ territorial organisation, with national and regional elites blaming each other for inadequate containment responses and demanding centralisation or decentralisation in response (Amat and Rodon 2023). The increased salience of many territorial questions in this context gave rise to a renewed discussion about which territorial authority structures would best deliver effective policy responses in times of crisis (OECD 2020). Ultimately, the crisis context gave national governments the power to invoke emergency laws that allow for a concentration of power at the national level and a radical shift in the territorial structure of authority.

Interestingly, however, countries varied widely in how they allocated authority across levels of government. Some countries, such as France, opted to centralise authority for protective policies at the national level. Other countries, like the United States, took a decentralised approach, leaving authority over the containment response in the hands of subnational governments. Yet another set of countries changed their authority arrangement midway through the first year of the pandemic. How should we conceive of this variation?

This paper contributes to the emerging literature on territorial authority in crisis contexts in two important ways. First, it proposes an inclusive measurement scheme of how authority over protective policies was allocated in the context of the pandemic, which we subsequently use to assess 13 advanced democracies. Relevant policies include school closures, restrictions of business activities, cancellations of public events and gatherings, public transport closures, stay-at-home orders, and travel restrictions (see Hale et al. 2021). The proposed measure differs from existing data collection projects in that we capture how authority over protective measures was allocated across levels of government, rather than the stringency of these measures.¹ We code constitutional documents and specific COVID-19 legislation complemented with press releases by national and regional governments and media articles.² Expanding the study beyond the small number of highly decentralised cases that have formed the focus of much of the recent work allows for a more systematic comparison of territorial authority arrangements during the crisis. The measure distinguishes between two different dimensions of a country’s territorial authority arrangement during the pandemic: (1) the extent to which regional governments, rather than the national government, have authoritative competence over protective policies in their own territory, with full authority signifying a high level of decentralisation, and (2) the institutions through which national and regional governments share authority over protective policies.

The paper’s second contribution is that it tracks changes in the territorial structure over the first 12 months of the outbreak, which allows us to explore whether territorial arrangements were responsive to the changing nature of the pandemic. We begin our observations shortly before the pandemic in January 2020 to assess whether the outbreak led to a change in territorial arrangements. We end our observations in December 2020, after which vaccination programmes slowly started to be rolled out, and protective policy measures began to lose their importance as the only meaningful policy tool to contain the spread of the pandemic.

Our time-variant measure allows us to offer new descriptive insights into the territorial structure of authority during the crisis, which we believe to be of intrinsic value. We can summarise four main patterns: (i) (De)centralisation efforts during the first year of the pandemic follow an up-and-down pattern prescribed by the urgency of the crisis. There was a strong tendency to centralise authority over protective policies during the first wave of the pandemic, which ebbed during the summer of 2020 when COVID case numbers decreased, and increased again during the second wave of the pandemic without, however, reaching first-wave levels; (ii) Countries where the centralisation of authority over protective policies is constitutionally not possible tended to rely on institutional incentives for coordination as a functional alternative; (iii) The broader territorial arrangements in a country influenced the allocation of authority over protective policies during the pandemic; (iv) Where the national crisis response was politicised, and regional crisis competence was strong, regional governments were likely to obstruct the national government’s centralisation and coordination attempts.

These descriptive insights contribute to our understanding of the extent to which the territorial allocation of crisis competencies was rooted in a country’s general territorial arrangements, but they also highlight the potential for rapid change in the context of a crisis. They raise several interesting questions, which we use as starting points for theory generation, such as: To what extent were territorial arrangements over protective policies influenced by functional pressures for rapid crisis response? Is shared competence over protective policies a functional alternative to centralisation to the degree that it delivers similarly effective results? And are low levels of partisan-political contestation between regional and central government a precondition for centralisation and power sharing? In providing tentative answers to these questions, we engage in some inductive reasoning about the factors explaining the cross-national and temporal variation in territorial arrangements over protective policies.

We also specify the scope conditions under which we believe these patterns might travel to other crisis contexts. These conditions include a highly interdependent context and a crisis that impacts a sizeable share of a country’s territory such that it cannot be isolated or dealt with bilaterally between an individual region and the national government.

The paper proceeds as follows: We begin by exploring the existing measures of decentralisation and the extent to which they are applicable to the territorial structure of authority during the crisis. We then introduce our own measure of authority allocation during the crisis by using illustrative cases to highlight coding issues. In the last section, we apply the scheme to 13 advanced democracies over the first 12 months of the pandemic and lay out our four descriptive patterns. In conclusion, we address the scope conditions under which we believe our findings might travel to other crisis contexts.

Two Dimensions of Territorial Arrangements

Emerging case study research on COVID-19 suggests that countries varied widely in how they allocated authority over protective policies across levels of government, with some countries using emergency laws to centralise authority and others letting regional governments manage protective policies. However, countries did not only vary in where authority was located, i.e. in how (de)centralised their crisis response was. They also varied in the way in which different levels of government related to one another. More precisely, in some countries, such as Germany, regional governments interacted frequently with the national government in the context of power-sharing institutions. Consensus building within these institutions shaped the overwhelming share of protective policies (Vampa 2021). In countries like the United States, by contrast, state governments acted more autonomously and institutional incentives for greater interaction were largely absent (Lecours et al. 2021; Vampa 2021). The decentralisation of authority coupled with few institutional incentives for more and better intergovernmental interactions prevented any meaningful policy coordination. The conflictual style of the Trump presidency and intense partisan competition further contributed to acrimonious relations between levels of government.

The idea that the extent to which authority is decentralised and the way in which levels relate are two qualitatively different aspects is consistent with the existing theoretical and empirical literature on territorial governance (Elazar 1991; Bednar 2009; Hooghe et al. 2010; Bolleyer and Thorlackson 2012). It has now become accepted practice to disaggregate measures of territorial arrangements of into two separate dimensions: the degree to which authority has been decentralised to regional governments, also known as self-rule, and the degree of power sharing, known as shared rule. The terms self-rule and shared rule draw on Elazar’s seminal work on federal systems (Elazar 1991). Self-rule is sometimes measured using the share of subnational expenditures and revenue as a proxy but more recently measures have sought to capture the policy competencies allocated to regional governments directly by coding constitutional documents (Arzaghi and Henderson 2005; Brancati 2006; Bolleyer and Torlakson 2012; Hooghe et al. 2016). Shared rule is often captured using the strength of various power-sharing institutions as a proxy (Brancati 2006; Hooghe et al. 2010, 2016). Power-sharing institutions include a strong second legislative chamber that represents regional interests in the national law-making process and institutionalised executive-level meetings between national and regional representatives (Hooghe et al. 2016). Some scholars have ventured to measure power sharing between levels of government by focusing on legislative interdependence. Legislative interdependence describes the extent to which legislation is shared by both levels, regional and national, rather than being exclusively assigned to one level (Bolleyer and Thorlakson 2012). The two approaches may be more consistent with one another than apparent at first sight. This is because the interdependence that arises from shared legislative responsibilities or other forms of complex responsibility assignments often generates conflict between levels of government, which tends to necessitate more elaborate institutional procedures in the form of power-sharing institutions (Bednar 2009; Bolleyer and Thorlakson 2012).

However, existing two-dimensional measures are somewhat limited in their applicability to the crisis context for two reasons. First, existing measures capture a country’s general territorial structure but are insensitive to variation across policy areas. Empirical evidence shows that policy areas are decentralised to varying degrees (Schakel 2009; Ostrom 2010; Dardanelli et al. 2019). Applying general measures of territorial authority to specific policy areas may therefore lead to incorrect classifications. Protective policies, and public health policies more generally, are particularly difficult to capture with generalist measures of territorial authority. Authority over public health typically lies with regional or even local governments but it is frequently subject to emergency laws that allow the national governments to reign in power rapidly in a pandemic. The added complexity of emergency legislation and the potential for drastic changes in the territorial structure of authority makes protective policies difficult to capture using existing generalist indices of territorial authority.

Secondly, existing measures typically assess territorial structures at yearly time intervals, which has proven sufficient for most purposes because territorial authority structures are relatively resistant to redesign (Hall and Taylor 1996; Pierson 2000; Hooghe et al. 2010). However, the pandemic challenged this conventional wisdom and brought to the fore the fragility of these territorial arrangements. The unprecedented shock it unleashed on countries’ territorial structures opened opportunities for drastic changes to countries’ territorial organisation, including the wave towards centralisation during the first few weeks of the pandemic and its reversal in the summer of 2020. A monthly measure is, therefore, better suited to tracking these changes and understanding how territorial structures respond to different phases of the pandemic.

Measuring the Territorial Structure of Authority

Having highlighted the need for a new measure of territorial authority in the context of the pandemic, we now lay out our coding scheme. The scheme operationalises territorial authority over protective policies in a way that is consistent with the two-dimensional conceptualisation of territorial organisation established in the existing literature (Hooghe et al. 2010, 2016). Our measure is, by definition, narrower and more policy-specific than more generalist measures of territorial authority, like the Regional Authority Index (RAI). What we propose is a COVID-19-specific application of the self-rule and shared-rule dimensions of the Regional Authority Index. We refer to self-rule and shared rule as regional crisis competence and shared crisis competence, respectively, to highlight that this measure differs from more general measures.

We conceive of regional crisis competence as the degree to which authoritative competence over protective policies has been allocated to regional governments rather than the central government. The more competencies are allocated to the regional government, the more decentralised the structure, and vice versa. The fewer competencies are allocated to the region, the more centralised the structure. We conceive of shared crisis competence as the degree to which the national and regional governments share competencies over protective policies. This dimension ranges from a strict separation of competencies over protective policies to shared crisis responsibility (see Bolleyer and Thorlakson 2012). Crucially, shared crisis competence refers to a particular type of authority structure and should not be equated with collaborative decision-making. Shared competences may incentivise collaboration, but they do not guarantee collaborative outcomes. The coding scheme captures formal competencies over protective policies that are exercised following explicit rules, which are usually but not always written in constitutions and legislation.

Coding schemes face the difficult task of having to identify common rubrics based on commonalities across observations, which need to be both general enough to transcend the individual case and specific enough to clearly classify individual cases (Hooghe et al. 2010). While setting out the coding scheme, we will use illustrative cases to highlight some of these issues. For full transparency and replicability, the online appendix includes the individual coding profiles of each country and references the sources based on which coding decision was made. To increase the accuracy of our scores, we assigned two researchers to code the same country case and compare results on a one-to-one basis.

Regional Crisis Competence

To evaluate how much authority over protective policies was decentralised to regional governments, we need to assess the extent to which the regional governments were able to make decisions on protective policies without national government involvement. We distinguish four categories of regional crisis competence (Table 1).

Table 1. Regional crisis competence

0	Regional governments do not have the authoritative competence to increase or decrease the stringency of national protective policies.
1	Regional governments have the authoritative competence to increase, but not decrease, the stringency of national protective policies in one of the following areas: events and gatherings, business activity, public transport, and schools.
2	Regional governments have the authoritative competence to increase the stringency of national protective policies, or to decrease the stringency with prior authorisation, in at least two of the following areas: events and gatherings, business activity, and public transport, and schools.
3	Regional governments have the authoritative competence to introduce protective policies in all the following areas, with the national government merely offering guidance. events and gatherings business activity public transport schools and they are endowed with the authority to issue stay-at-home orders and travel restrictions.

The first is a null category where regional governments do not have the authoritative competence to increase or decrease the stringency of national protective policies. France for example receives a score of 0 starting on 14 February 2020, when the national government activated the national-level emergency plan.³

The second category encompasses cases where regional governments have the authoritative competence to increase, but not to decrease, the stringency of national protective policies in one of the following areas: events and gatherings, business activity, public transport, and schools. National measures set a strict minimum threshold. That is, regional governments may only increase the stringency of these measures, but not decrease it. In these cases, national policies overwhelmingly guide the crisis response and regional governments may only modify limited aspects.

The third category describes regional governments that have the authoritative competence to increase the stringency of national protective policies, or to decrease the stringency with prior authorisation, in at least two of the following areas: events and gatherings, business activity, public transport, and schools. Once again, national governments are more concerned about establishing a minimum threshold and therefore make it more difficult to decrease the stringency of national measures.

The difference between a score of 1 and 2 is often a matter of degree and depends on how many aspects of the crisis response regional governments can modify. What is more, regional governments may only increase the stringency of protective policies when countries score a 1, but they may increase or decrease the stringency when countries score a 2. They do, however, need prior authorisation when decreasing the stringency. The case of Austria best illustrates these distinctions. On 10 March 2020, the Austrian federal government closed shops and restaurants, banned large outdoor and indoor events, made face masks mandatory, and introduced remote learning at universities (Bundeskanzleramt, 2020a). The Länder governments mostly implemented the national government’s measures, but they also had the possibility of increasing the stringency of restrictions on business activities, which is why Austria receives a score of 1 during this period. On 15 May 2020 national government restrictions were eased and Länder governments had the option of increasing the stringency of restrictions across a wide range of areas, including school closures and curfews (Bundeskanzleramt, 2020b). Austria therefore receives a score of 2 during this period. The case of Austria also nicely illustrates how our scoring schema is responsive to changes over time.⁴

The final category encompasses cases where regional governments have the authoritative competence to introduce protective policies in almost every area, including events and gatherings, business activity, public transport, and schools. Regional governments also have the authority to issue stay-at-home orders and travel restrictions. In these countries, the national government’s primary role is limited to offering guidance and coordination in addition to regulating a few areas exclusively assigned to the national level, such as closing national borders and regulating international travel. In these countries, the right of regional governments to respond to a public health crisis is typically rooted in the constitution, as is the case in Australia, Canada, Germany, and the United States. In the United States, for example, the 10th Amendment of the US Constitution has been interpreted as granting states the exclusive authority to respond to public health crises. The United States, therefore, scores a maximum of 3 points (Library of Congress 2020).⁵

Shared Crisis Competence

In line with existing measurement approaches, we focus on the presence and strength of power-sharing institutions to assess the extent to which national and regional governments shared crisis competence and jointly decided on protective policies (Hooghe et al. 2010). Because the use of emergency legislation during the pandemic shifted the balance of power towards executive decision-making without parliamentary input and little parliamentary oversight, we focus on executive power-sharing institutions in the form of intergovernmental meetings. These are more or less regular meetings of members of the executive branch of government, comprising either first ministers or line ministers (Behnke and Mueller 2017). We distinguish three categories (Table 2). To score more than a 0, meetings between the national government and regional representatives must be more than purely informational. To receive a score of 2, meetings between the national government and regional governments must reach decisions that are legally binding on the participants.

Table 2. Shared crisis competence

0:	No meetings between the national government and regional governments concerning protective policies or purely informational meetings.
1:	Frequent meetings between the national government and regional governments concerning protective policies that are consultative without legally binding authority.
2:	Frequent meetings between the national government and regional governments concerning protective policies with authority to reach legally binding decisions.

The Swiss case best illustrates the fundamental distinction between the first and second category. The 2012 Epidemics Act, which provides the legal basis for Switzerland’s COVID-19 response, differentiates between three different alert levels. Under so-called normal circumstances, cantonal governments have responsibility for the policy response. Under exceptional circumstances, the national government can impose protective policies after consulting with cantonal governments. These consultative meetings are obligatory, but they do not legally bind the national government. Under so-called extraordinary circumstances, which were declared on 16 March 2020, the national government can impose measures without consulting the cantonal governments first (Epidemiengesetz 2012). During the “extraordinary” period, which lasted until mid-June, Switzerland scores a 0 (Schweizer Bundesrat 2020). When the alert level was reduced to exceptional circumstances in mid-June 2020, the national government did resume consultations with the cantonal governments, thus meriting a score of 1.⁶

To score the maximum 2 points on this scale, meetings between the national government and regional governments must reach decisions that formally bind the participants. These decisions are typically reached by consensus, not majority rule, which may lead to lengthy decision-making processes. In Australia, for example, the National Cabinet, created on 13 March 2020, quickly absorbed the function of peak intergovernmental forum of the previous and now-defunct Council of Australian Governments (COAG). It is comprised of the Australian Prime Minister and the heads of states and territories, meets on a near-daily basis, and creates legally binding rules (Morrison 2020). Australia, therefore, scores the maximum value of 2. Across our sample, scoring a 2 on shared crisis competence measures is associated with a high degree of regional crisis competence. This means that the central government cannot simply impose a decision in the absence of a mutual agreement by consensus.

A comparison between our newly developed measure and the widely used Regional Authority Index is useful to test its validity but also to help highlight some of the differences between the measures. Most importantly, the RAI captures a country’s general territorial arrangement and includes a much broader range of aspects including fiscal and electoral features, while our proposed measure is narrower and focuses on protective policies (Hooghe et al. 2010). However, the RAI is consistent with the construction of our measure to the degree that it also distinguishes between two dimensions of territorial authority: the allocation of authority between the regional and the national level, referred to as self-rule, and the extent to which regional governments share power over the territory as a whole in the context of power-sharing institutions, referred to as shared rule (see Hooghe et al. 2016). Using a principal factor analysis we compare the latest available country-level self-rule and shared-rule scores from the RAI with our respective measures for regional crisis competence and shared crisis competence, which reveals that there are commonalities between the different indexes.⁷ In each of the two comparisons the principal component has an eigenvalue well above 1, indexes correlate by more than 0.7, which points to a common underlying structure.⁸

Analysis

We now apply the measurement instrument developed in the previous section to evaluate the territorial structure of authority during the COVID-19 crisis across countries and over time. We choose 1 January 2020, just before the onset of the pandemic, as the starting point for our analysis. Choosing to start our observations before the actual onset of the crisis might seem counterintuitive. However, because our measure captures competencies over protective policies, rather than the stringency of these protective policies, it is possible to establish a January 2020 baseline measure. This is because competencies over protective policies are codified in constitutions or specific public health legislation, both of which were in place before the pandemic. The baseline measure allows us to discern whether newly passed emergency legislation caused a noticeable change in the territorial authority over protective policies compared to before the pandemic. We end our observations at the close of the first year of the pandemic on 31 December 2020, which gives us 12 data points for each country. This is a natural endpoint for the analysis, as our measures focus on protective measures. With the rollout of vaccination programmes in 2021, protective policies started to lose their relevance as the main tool in effectively combating the spread of the virus.

The descriptive analysis is deliberately restricted to 13 stable democracies that have a defined regional level of government.⁹ First, an examination of shifts in the allocation of authority from regional governments to the national government makes having a defined regional level a requirement for selection. Second, established democracies and the associated characteristics of constitutionalism, the rule of law, and judicial constraints mean that the use of emergency laws and ensuing centralisation of power is not likely to be subject to opportunistic power grabs of authoritarian rulers. Shifts in the allocation of authority can, therefore, reasonably be attributed to the pandemic itself.

The list of countries included in this analysis is not exhaustive but represents a smaller and logistically feasible subset of the universe of cases that are established democracies with a defined regional level. In line with recent discussions that call for more openness about the logistical considerations that go into case selection (see Koivu and Hinze 2017), we want to be transparent about the fact that limited access to government documents, language skills, and funding limitations constrained our case selection. The subset of 13 countries reflects the difficult task of weighing these constraints against our research goals while seeking to preserve methodological rigour. While the list of countries is not exhaustive, it is reflective of the wide variety of institutional set-ups and includes established federal countries as well as countries that have decentralised authority more recently.

The measure we developed lends itself to being used for multiple purposes, including for causal inference, but the primary aim of this paper is to arrive at new descriptive insights, an often neglected and intrinsically valuable area of political science (Gerring 2012). Our descriptive analysis of territorial authority structures during the crisis reveals four main patterns, which we detail below. These descriptive patterns raise several interesting questions: Were changes in the territorial structure of authority driven by functional pressures, such as the need for rapid decision-making? And how did the broader institutional framework influence the direction and intensity of these changes? They also highlight the degree to which the crisis response was subject to party-political contestation, which leads to questions about how the political context might have shaped territorial arrangements (Lecours et al. 2021). While we cannot answer these questions conclusively in the context of this paper, we will engage in some theory generation, but not theory testing, in association with these descriptive patterns and point to factors that might help explain the cross-national and temporal variation in the allocation of authority over containment measures during the COVID pandemic.

Findings

Efforts to Centralise Authority, by Restricting Regional Crisis Competence, Mirror Changes in the Urgency and Unpredictability of the Crisis

Our time-variant measure for regional crisis competence reveals a notable trend towards centralising authority and restricting regional crisis competence over protective policies during the height of the first wave of the pandemic in April 2020. Figure 1 represents the monthly changes in regional crisis competence for all the countries in our dataset, with the x-axis denoting different months, starting in January 2020, right before the pandemic. It shows that 8 of the 13 countries in our dataset curbed regional governments’ crisis competence and shifted authority to the national government during the first wave. This finding is broadly in line with existing case study research, which also observes a trend towards centralisation in the spring of 2020 (Greer et al. 2022). Interestingly, five out of these countries – Austria, Belgium, Italy, Spain, and Switzerland – reinstated regional governments’ authority over protective policies during the summer of 2020, when COVID-19 case numbers declined, and the sense of urgency decreased. However, with the start of the second wave in autumn 2020, three of those five countries – Austria, Belgium, and Italy – once again restricted regional governments’ crisis competence, albeit not to the same extent as in the first phase of the pandemic.

Figure 1. Monthly changes in regional crisis competence by country and month in 2020

This up-and-down pattern, with changes in regional crisis competence mirroring changes in the urgency of the crisis, suggests that functional pressures may indeed be an important factor influencing territorial arrangements. A functional advantage of centralised systems is the ability to rapidly develop an appropriate policy response in unprecedented situations (Kincaid et al. 2010). The fact that centralisation during the second wave of the pandemic did not reach first-wave levels further supports this notion, as the crisis lost some of its unpredictability and developed into a routine emergency.

The pattern is also compatible with theoretical work suggesting that centralised systems are better able to internalise policy spillovers (Oates 2005). Policy spillovers are negative (or positive) externalities in highly interdependent contexts, where one territorial unit’s policy response is likely to immediately affect the other units. An often-discussed policy spillover during the pandemic, especially in the US context, arose from the unilateral decision by some regions to lift or decrease restrictions, which resulted in a patchwork of protective policies and incentivised cross-border travel. This, in turn, created new hot spots and undercut the efficiency of existing measures. Similar to approaches that stress the superior ability of centralised systems to arrive at rapid decisions, research on policy spillovers focuses on functionality as an important driver of territorial governance structures.

However, 5 of the 13 countries in our dataset saw no change in regional crisis competence.¹⁰ They maintained a highly decentralised response throughout the entire year, which points to a more complex mechanism by which functional pressures influence governance structures.

Shared Crisis Competence May Act as a Functional Alternative to Centralisation

Our measure of territorial authority also reveals changes in the level of shared crisis competence throughout the pandemic, though less drastic than in the case of regional crisis competence (Figure 2). Only 5 of the 13 countries constrained the degree of shared crisis competence during the first wave of the pandemic. The number of countries experiencing changes in shared crisis competence is therefore lower than the number of countries that experienced changes in regional crisis competence.

Figure 2. Monthly changes in shared crisis competence by country and month in 2020

In countries that did constrain shared crisis competence, changes tended to involve somewhat more moderate shifts, and in three of these countries – Italy, Switzerland, and Spain – it was reinstalled at the end of summer 2020.¹¹ At first glance, then, it appears as if shared crisis competence was more resistant to change than regional crisis competence, which would be in line with research suggesting that power-sharing institutions are indeed difficult to reform, even in the face of crisis (Marks et al. 2008).

To understand how the two dimensions of our measure of territorial authority relate to one another, we have classified each of our cases according to how their regional crisis competence score and their shared crisis competence score in the bivariate table below (Table 3). Each cell in the table shows a particular combination of regional and shared crisis competence and the empirical cases that represent that combination. Some cases only exhibit one combination throughout the pandemic, whereas some countries change from one cell to another. Cases in black represent the combination during the first wave of the pandemic, whereas cases in light grey represent combinations in the summer of 2020.

Table 3. Regional crisis competence and shared crisis competence

		Shared crisis competence
		0	1	2
Regional crisis competence	0	Centralised France Japan Norway Italy Spain Switzerland
	1		Centralised consultative Austria Belgium
	2		Decentralised consultative Austria Belgium Italy Switzerland	Decentralised coordinated Spain
	3	Decentralised Canada United States UK	Decentralised consultative UK	Decentralised coordinated Australia Germany

This bivariate table reveals an interesting clustering of countries. The cluster in the top-left combines low levels of regional crisis competence with non-existent shared competence. In these cases, authority over protective policies was in the hands of the national government and regional governments were not involved in the decision-making. We conceive of the first combination as a “centralised” structure.

The cluster in the bottom-right corner, diametrically opposed to the first cluster, refers to countries where high regional crisis competence was combined with a high degree of shared crisis competence. This cluster notably includes Australia and Germany, where strong power-sharing institutions in the form of executive-level intergovernmental meetings were in place or were specifically created to deal with crisis containment. Crucial decisions regarding protective policies were made in the context of these institutions without, however, impinging on regional crisis competence (see online appendix). Compliance with containment decisions made in the context of these institutions was high, although it waned somewhat during the later stages of the pandemic, especially in the case of Germany. We refer to this as a “decentralised coordinated approach”. Recent case studies suggest that this set-up was highly successful in incentivising coordination across the level of government, which resulted in an effective containment response on a par with some of the more centralised approaches (Vampa 2021).

The prediction that shared crisis competence is a functional alternative to centralisation to the degree that it can deliver a similarly effective crisis response is consistent with the recent literature on the policy implications of power-sharing institutions. This literature shows that power sharing can promote coordination across levels of government and constrain some of the centrifugal aspects associated with high degrees of regional self-rule, which results in a more coordinated and homogenous policy (Kleider 2018). Compared to a centralised approach, shared crisis competencies may not deliver equally rapid responses, but it may have the advantage of regional governments being able to respond to varying contexts and engage in a dialogue with local citizens, both of which should positively impact the effectiveness of containment measures (Scholz et al. 1991). Last but not least, shared crisis competence may also be a more conservative and less disruptive intervention than restricting regional crisis competence and might therefore be less likely to invite obstruction or regional mobilisation, especially in highly politicised contexts.

The third cluster of countries is characterised by very high levels of regional crisis competence and no meaningful shared crisis competence, which necessarily leaves large scope for unilateral action on the side of regional governments. We refer to this type as a “decentralised approach”. This cluster includes the United States and Canada.

Between these three extremes, there are various combinations of regional crisis competence and shared crisis competence. For instance, we observe the combination of low levels of regional crisis competence with consultative intergovernmental meetings in Austria and Belgium, which we refer to as the “centralised consultative” structure. Intermediate levels of regional crisis competence combined with consultative intergovernmental meetings occur in Austria, Belgium, Italy, and Switzerland during the summer, which we classify as a “decentralised consultative” approach to crisis management. Spain constitutes an interesting case of institutional innovation. Its intergovernmental public health body CISNS (Consejo Interterritorial del Sistema Nacional de Salud) comprised of the national health minister and the health councillors of each of the 17 autonomous communities, evolved from playing a very limited role at the outset of the pandemic to issuing formal opinions on protective policies by the summer of 2020 which informed the content of executive decrees. We therefore classify Spain as a “decentralised coordinated” system starting in the summer of 2020. While all combinations of regional and shared crisis competence are theoretically possible, there are no empirical examples of low degrees of regional crisis competence – i.e. high levels of centralisation – combined with shared crisis competence. This is because shared crisis competence requires a minimum degree of regional competence to be viable.

The Allocation of Authority over Protective Policies Is Rooted in Countries’ Broader Territorial Structure

For a fuller understanding of the territorial organisation of protective policies during the pandemic, it might be useful not to view it in isolation but rather against the backdrop of a country’s broader territorial structure. To this end, we compare our newly developed measure with one of the most widely used general measures of territorial arrangements, the Regional Authority Index. The RAI is consistent with the construction of our measures as it also disaggregates territorial authority into two dimensions: self-rule and shared rule. However, the RAI incorporates a much wider variety of aspects than our policy-specific measure and is, therefore, better suited to capturing a country’s general territorial structure (Hooghe et al. 2016).¹²

Figures 3 and 4 plot the latest available data for the RAI dating from 2018 against our policy-specific measure. Figure 3 shows a relatively strong correlation between the RAI’s self-rule dimension and regional crisis competence before the onset of the pandemic. However, during the height of the first wave in April 2020, this correlation weakened. This is because emergency laws invoked during the first wave of the pandemic shifted authority over protective policies from the regions to the national government, even in countries where regional self-rule is otherwise strong, like Spain and Switzerland.¹³

Figure 3. Regional crisis competence in January 2020 and April 2020

Figure 4. Shared crisis competence in January 2020 and April 2020

These findings provide mixed evidence for path-dependency arguments. While territorial arrangements over protective policies appear to be fairly rooted within countries’ broader territorial frameworks, they are also changeable in the event of a crisis, especially when they are subject to emergency legislation. Another interesting observation is that countries with low degrees of self-rule – Japan, France, Norway – experienced somewhat more muted and predictable changes in regional crisis competence. In those cases, already low baseline levels of regional crisis competence were further restricted to zero. By contrast, there is significant variation in the extent to which countries with high levels of self-rule responded to the pandemic. While regional crisis competence barely changed in countries like Australia, Canada, Germany, and the US, it underwent drastic changes in countries like Austria, Belgium, Italy, Spain, and Switzerland.

Compared to the association between regional crisis competence and the RAI’s self-rule dimension, the relationship between shared crisis competence and the RAI’s shared-rule dimension appears more robust (see Figure 4). This finding is broadly consistent with research which suggests that power-sharing institutions are difficult to set up and remove once in place, with reforms often requiring two-thirds majorities (Marks et al. 2008). However, the finding contradicts recent case studies that do not find an association between strong shared-rule practices before the pandemic and the use of power-sharing institutions during the crisis (Hegele and Schnabel 2021).

Regional Governments Are Likely to Reject the National Government’s Centralisation and Coordination Attempts when Partisan Political Competition between Levels of Government Is Intense

Canada and the United States are the only countries where regional crisis competence remained largely unconstrained and power-sharing institutions were weak or absent. Neither country invoked emergency legislation, and no power sharing in the form of regular meetings between the national government and regional governments took place.¹⁴ At first sight, this particular territorial arrangement appears to be at odds with a functionalist understanding as it reaps neither the benefits of rapid decision-making in centralised systems nor the benefits of a coordinated approach via power-sharing institutions. Exploring these divergent cases in more detail may therefore present a useful starting point for generating hypotheses about the conditions under which other influences may eclipse functional pressures.

Recent case studies point to the influence of political factors on territorial arrangements in both countries, specifically the extent of partisan competition between the national and regional levels combined with a highly politicised crisis response. Especially in the United States, this combination resulted in highly acrimonious intergovernmental relations further fuelled by the conflictual style of the Trump presidency, with the President and Democratic governors blaming each other for an inadequate containment response (Bennouna et al. 2021; Lecours et al. 2021). The lack of coordination created a patchwork of protective policies in the United States, with cross-border travel into states with fewer restrictions creating new Corona hotpots and further increasing the reproduction rate of the virus (Yong 2020). Because federal and provincial parties are not vertically integrated in Canada, Conservative provincial premiers had less interest in opposing measures proposed by the Liberal Trudeau government (Lecours et al. 2021). This resulted in somewhat less conflictual intergovernmental relations than in the US, aided by Trudeau’s unifying leadership approach in the early stages of the pandemic (Broschek 2022). Yet provincial premiers still successfully rejected several of Trudeau’s centralisation and coordination efforts, including his attempt to invoke the Emergencies Act in April 2020 (Cochrane et al. 2020).

In the absence of strong power-sharing institutions which could have provided the necessary focal point for coordinating regional protective policies, regional governments took matters into their own hands. In both countries, bottom-up attempts to coordinate policy responses emerged. For instance, in late April 2020, governors of several US states decided to share information, research, and resources and coordinate their policies. This resulted in the formation of three distinct “blocs”: the Eastern States Council, the Midwest Governors Regional Pact, and the Western States Pact formed by the states of California, Oregon, Washington, Colorado, and Nevada (Kelly and Sgueglia 2020; Coronavirus Resource Center 2021). In all three regional blocs, states agreed to a uniform policy on social distancing, although agreements were not legally binding (Greve and Koran 2020; Kelly and Sgueglia 2020; Bennouna et al. 2021). Bottom-up coordination among Canadian provinces was less developed and mostly involved inter-provincial travel regulations. The so-called Atlantic Provinces – Nova Scotia, New Brunswick, Prince Edward Island, and Newfoundland and Labrador – created the “Atlantic Bubble” on 24 June. It involved a quarantine-free zone for travellers within those provinces (Council of Atlantic Premiers 2020).

Conclusion

The COVID-19 pandemic unleashed a shock on countries’ territorial organisation and highlighted the potential for radical shifts in countries’ territorial arrangements in the context of crisis (Amat and Rodon 2023). Although most of these changes did not extend to post-pandemic times, they set a precedence for radical reorganisation during times of crisis. This paper proposes a new measure that allows us to track these changes during the pandemic and systematically compare countries’ territorial arrangements over protective policies. The measure, which captures regional crisis competence and shared crisis competence, has several advantages over existing measures. First, existing measures focus on a country’s general territorial organisation and do not capture the idiosyncrasies of specific policy areas, which may sometimes differ substantially from the general context. This is the case with public health, which eludes general measures of territorial organisation because it is often subject to specific pandemic laws or general emergency laws that allow the national governments to reign in powers rapidly in the event of a crisis. Second, existing measures are provided on a yearly basis and are therefore ill-suited to capturing the rapid changes in countries’ territorial arrangements that occurred during the crisis.

The new measure of territorial authority allows us to arrive at four novel descriptive insights, which we use as a starting point for theory generation. First, we find that efforts to centralise authority, i.e. restrict regional crisis competence, mirror the urgency and unpredictability of the crisis. Centralisation peaked during the first wave of the pandemic, eased during the summer of 2020 and rose again during the second wave in autumn 2020. This up-and-down pattern is consistent with some functional arguments that point to the advantage centralised systems have at responding rapidly and internalising negative policy spillovers. However, not all countries centralised authority and restricted regional crisis competence, which suggests a more complex mix of motivations influencing territorial arrangements during the crisis.

Second, we find that power-sharing institutions played an important role in delivering coordinated containment measures in some of the countries where regional crisis competence was not restricted, most notably Australia and Germany. Shared crisis competence, therefore, emerged as a highly effective alternative to centralisation during the first year of the pandemic to the degree that it delivered a similarly efficient and coordinated crisis response.

Third, the specific territorial structure of authority over protective policies during the crisis appears rooted in a country’s broader territorial organisation. However, the norms guiding shared crisis competence are more path-dependent than regional crisis competence. This is broadly in line with research showing that power-sharing institutions are highly path-dependent as they are immensely difficult to set up and remove once in place. Though clearly embedded in the broader territorial structure, the level of regional crisis competence appears to be somewhat more changeable in response to crisis contexts, with emergency laws enabling national governments to override regional crisis competence even in contexts where regional powers are otherwise strong, as in Switzerland.

Lastly, we find that partisan competition between regional and national governments, combined with a highly politicised crisis response, can lead regional governments to resist centralisation and coordination attempts initiated by the central government. Moderate or low levels of party-political contestation of the crisis response, therefore, emerge as an important condition under which functionalist pressures can reasonably be expected to lead to a centralisation of authority or to coordination via power-sharing institutions.

These findings make a substantial contribution to our understanding of cross-national and over-time variation in territorial arrangements during the pandemic. Still, the important question is whether they are generalisable to other crisis contests. It is tempting to assume that these findings apply to any catastrophe subject to emergency laws which can shift a country’s territorial structure. These typically include epidemics, major natural disasters, Chornobyl-scale industrial accidents, and lastly wars, for which most modern emergency legislation was originally devised. The common thread in these different crises is that they typically affect a sizeable share of a country’s territory, such that they cannot be dealt with in an isolated manner. A more localised crisis is typically dealt with bilaterally between an individual region and the national government so that the broad power shifts we describe would not necessarily apply. Hence the territorial scope of a crisis appears to be an important scope condition. Indeed, in the context of the COVID-19 pandemic, legislators frequently felt the need to highlight that it was an epidemic situation of national significance. A second and related scope condition is the level of interdependence, i.e. the extent to which the decisions of individual regional units can negatively – and sometimes positively – affect other units. We would expect that interdependence is related to the demand for constraining unilateral action by restricting regional crisis competence or coordinating regional decisions in the context of power-sharing institutions. For instance, it is easy to imagine how our findings might apply in the context of major environmental disasters, like the 2021 European floods, where individual governments’ actions related to floodgates impacted upstream and downstream regions. However, other crisis contexts might be less interdependent and create less need for coordination or centralisation. A final scope condition relates to regional competence in crisis-relevant policy areas. To the surprise of some pandemic observers, competence for protective policies and other areas of public health measures often lay with regional governments. This required recourse to emergency laws and/or specific pandemic legislation to provide a legal basis for shifting authority towards the central government, and it called for coordination in contexts where regional governments maintained the competence. However, regional governments may play a more marginal role in crisis contexts where the initial authority in relevant policy areas already lies with the national government, which would limit the applicability of some of our findings.

Acknowledgements

We owe special thanks to the numerous research assistants working on compiling and editing the country profiles: Erona Tasholli, Frida Koslowski, Alessandra Sciarra, Lucas Lopes Silva, and Mingyu Sin. The project was funded by the Leverhulme grant “Coordinating against the Coronavirus: What lessons can be learned from multi-level states policy responses to Covid-19?” and a BA/Leverhulme small research grant obtained in 2020.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Supplementary Material

Supplementary data for this article can be accessed at https://doi.org/10.1080/13876988.2024.2332937.

Additional information

Funding

This work was supported by the British Academy.

Notes on contributors

Hanna Kleider

Dr Hanna Kleider is an Associate Professor (Senior Lecturer) in the Department of Political Economy at King’s College London. She received a doctorate in political science from the University of North Carolina at Chapel Hill. Her research interests center on comparative political economy, with a particular focus on decentralization, federalism and social policy.

Simon Toubeau

Dr Simon Toubeau is an Assistant Professor in the School of Politics and International Relation at the University of Nottingham. He obtained a PhD in Political and Social Sciences from the European University Institute (EUI). His research focuses on British politics and European politics, in particular, matters relating to the demands of nationalist parties like the SNP; the launching of independence referendums or campaigns witnessed in Scotland and Catalonia; and the processes of constitutional reform that devolve powers to regional governments.

Notes

1. Two excellent data collection projects have captured the stringency of the public response to COVID-19: The Oxford COVID-19 Government Response Tracker (see Hale et al. 2021) and the Protective Policy Index (PPI) project (see Shvetsova et al. 2021)

2. For a full list of documents please see the individual country profiles in the online appendix.

3. For more details, please see online appendix.

4. For more details, please check the country profile in the online appendix.

5. For more details, please check the country profile in the online appendix.

6. For more details, please see online appendix.

7. Since our questions mainly examine the territorial arrangement of a country as a whole rather than the authority of individual regions, we choose the aggregate country scores of the RAI to compare our measures with rather than the scores of individual regions.

8. See online appendix for detailed results of the factor analysis.

9. Countries included in the dataset are Australia, Austria, Belgium, Canada, France, Germany, Italy, Japan, Norway, Spain, Switzerland, the United Kingdom, and the United States.

10. The United Kingdom stands out as the only non-federal country with a highly decentralised response during the pandemic. The unusually high level of decentralisation in the UK’s crisis response is partially explained Boris Johnson allowing the devolved regions significant decision-making leeway (see online appendix).

11. The UK stands out as the only country that moved towards less shared rule over the course of the pandemic (see online appendix for more details).

12. The Regional Authority Index measures authority at the level of individual regions but also provides scores aggregated to the country which we use in our comparison. For an insightful analysis of individual regions during the pandemic see Shvetsova et al. (2021).

13. The UK is somewhat of an outlier because the devolved governments have a much higher degree of regional crisis competence than the RAI’s self-rule score suggests. This is due to the so-called “Sewel Convention”, which guided Westminster’s interactions with the devolved regions in the particular context of containment. The convention obliges the national government to seek the devolved regions’ agreement on all matters that directly affect them (Paun et al. 2022).

14. Canada’s Emergencies Act was not invoked (Cochrane et al. 2020). In the United States, the extent to which the President would have even had the legal authority to issue a national stay-at-home order remains hotly debated.