COVID-19 Policy Response and the Rise of the Sub-National Governments.

We examine the roles of sub-national and national governments in Canada and the United States vis-à-vis the protective public health response in the onset phase of the global coronavirus disease 2019 (COVID-19) pandemic. This period was characterized in both countries by incomplete information as well as by uncertainty regarding which level of government should be responsible for which policies. The crisis represents an opportunity to study how national and sub-national governments respond to such policy challenges. In this article, we present a unique dataset that catalogues the policy responses of US states and Canadian provinces as well as those of the respective federal governments: the Protective Policy Index (PPI). We then compare the United States and Canada along several dimensions, including the absolute values of sub-national levels of the index relative to the total protections enjoyed by citizens, the relationship between early threat (as measured by the mortality rate near the start of the public health crisis) and the evolution of the PPI, and finally the institutional and legislative origins of the protective health policies. We find that the sub-national contribution to policy is more important for both the United States and Canada than are their national-level policies, and it is unrelated in scope to our early threat measure. We also show that the institutional origin of the policies as evidenced by the COVID-19 response differs greatly between the two countries and has implications for the evolution of federalism in each. © Canadian Public Policy / Analyse de politiques.

Introduction

The outbreak of a novel coronavirus (coronavirus disease 2019, or COVID-19) at the beginning of 2020 provides a unique opportunity to study how different federations behave at the onset of a crisis. The resulting COVID-19 pandemic created conditions of high uncertainty and imminent threat and visibly affected how governments responded at different levels in both the United States and Canada. In responding to the threat, different levels of government were much more active than normal and often acted out of turn. For the first months of the pandemic, at least, these changes reorganized the ways in which policy-making processes are conducted at the national, state, provincial, and municipal levels. A stable federal balance between national and sub-national units persists in successful democratic federations such as the United States and Canada, although shifts, usually toward the relative strengthening of the federal level, take place slowly, if at all (Dardanelli et al. 2019). Yet, the COVID-19 pandemic may have tilted the policy-making activism toward the sub-national government in a matter of weeks and might prove to be the ultimate institutional un-balancing, or perhaps re-balancing, event.

In this article, we present an original dataset on the institutional origins of national and sub-national public health policy response during the onset phase of the COVID-19 crisis—the Protective Policy Index (PPI). We demarcate the onset phase as the period between 24 January (first diagnosed COVID-19 case in France, indicating that the virus was not contained) and 24 April 2020 (which was the last day in the period of non-decreasing sub-national policy response in both Canada and the United States).[1]

We then present a range of comparisons between Canada and the United States. The first dimension of comparison is the relationship between federal and sub-national versions of the index compared with the total protection enjoyed by citizens. We then examine the relationship between the evolution of the index over time and early mortality statistics by jurisdiction. Finally, we compare the institutional or governmental origin of the policies.

Our findings suggest that provinces and states—in particular, governors in the United States—have taken a convincing lead in developing and adopting policies to mitigate the COVID-19 pandemic, although the origins of these sub-national policies differ greatly between the two countries. We find no relationship in either federation between early threat to units as assessed by COVID-19 mortality in March 2020 and the speed or stringency of sub-nationally enacted policies. We do see substantial differences in the agency origins of public health response and conjecture that these differences are due to institutional variation. We conclude by providing several possible causes for these patterns and suggesting avenues for further research.

COVID-19 Protective Policy Index

On the basis of public health policy responses to COVID-19 at the provincial, state, and federal levels, we created the public health PPI, calculating the index for each jurisdiction on each day to indicate the level of policy protection of a resident in that jurisdiction (the unit of analysis is thus the unit-day). Using PPI data, we separately calculate provincial, state, and federal versions of the PPI. At present, the data do not contain information on municipal response, but we plan to include such information in the future.

We identify and code policies primarily from government resources, press releases, and reputable news sources, dating them on the basis of first announcement. Policies fall into several categories: border closures (international and domestic), school closures, social gathering and social distancing limitations, home-bound policies (curfew, stay at home, lockdown), medical isolation policies (self-isolation and mandatory quarantine), closure and restriction of businesses and services (closure of nonessential businesses, restaurants, entertainment venues, government offices, and public transportation; work-from-home requirements), the introduction of the state of emergency, and requiring mandatory personal protection equipment. The specifics of the index’s construction and the detailed discussion of the values assigned to categories and policies in it can be found in Appendix C in the online supplementary materials, and the codebook for the dataset is in online Appendix D. Although we did collect data on policies that specifically allocated funds for the purchase of protective equipment, ventilators, and other medical supplies, for the present purposes these policies are not included in the construction of the PPI.[2] Moreover, because the index is meant to measure protective public health policies, we do not include health system–related measures or anything related to health systems preparations that went on during this time. Although these are important, differences in the health system models of the United States and Canada make these comparisons difficult. We focus on measurable sub-national and federal public health policy response during the onset of the COVID-19 crisis.

The PPI was constructed by adding together the highest values in each category of the coded sub-national and federal policies on that day. Its daily possible minimum is zero, and its maximum is 40. Sub-national and federal PPIs were constructed with the values in each category from just state or just federal policies. It often happens that state and federal policies are duplicates of one another at a given time. For example, there may be a federally mandated self-isolation period and a self-isolation or quarantine requirement for a state or province based on the same or similar criteria. In this case, the federal policy would be counted toward the value of the federal PPI, and the sub-national policy would be counted toward the state or province PPI. The policy would only be counted once, however, for the total value of the PPI for an individual living in a given jurisdiction, for example for an individual in state or province j at time t: We have conducted consultations with public health practitioners as part of the content validation (Miller 2007) of these measures to ensure that we weighted the policies and categories according to their perceived effectiveness at the time.

It is important to note that, particularly in the United States, municipalities were often the first to move on COVID-19 policy. The true value of the PPI will thus be understated in this analysis for states and provinces with stronger municipal responses. We hope to supplement and expand this analysis once the municipal data are added to the data used in this work. Because our focus here is specifically the balance between national and sub-national policy responses in the COVID-19 emergency, we proceed with federal and state–province policy data.

Evolution of Policies

In the United States’ COVID-19 policy response, the first action tended to come from state and local governments, although governments at every level had limited ex ante information on what types of policies might be effective. Indeed, municipal authorities in major cities and counties that saw the first outbreaks became first policy responders, ahead of states (e.g., Los Angeles County, California; Houston, Texas; Cook County, Washington; Boston, Massachusetts; Mobile, Alabama; San Jose, California; San Francisco, California; Seattle, Washington). State governments eventually took the lead on COVID-19 policy response. By the time President Trump issued a state of emergency on 13 March 2020, 64 percent of the states had already declared a state of emergency and initiated their own policy responses. Action came from states spanning the partisan continuum: solidly Democratic California and Hawaii as well as the Republican strongholds of Utah and Indiana each issued their state of emergency declarations a week or more before the president did. Unlike in many past emergency responses, the national government generally acted after the states, endorsing non-binding guidelines put forth by the Centers for Disease Control and Prevention (CDC). These guidelines served as a template for the states that were initially reluctant to take action but did not become an enforceable policy in their own right, nor did they present as binding constraints in most states.

In Canada, the policy response was spread out between bureaucratic institutional response by health authorities (which in the United States was mostly absent) and provincial or federal government responses. Health Canada and the Public Health Agency of Canada (PHAC) have been at the forefront of Canada’s discourse and public education about the virus, with separate briefings being held by Prime Minister Justin Trudeau and Chief Public Health Officer Theresa Tam. The majority of actual COVID-19–related policy-making, however, has taken place within individual provinces. Moreover, within provinces, directives have come from provincial ministers of health and chief health officers and not through specific orders in council (roughly speaking in this context, the parliamentary counterpart to an executive order). Indeed, the main role of the latter (provincial states of emergency are enacted through order in council) seems to have been to enable the former to enact binding policies. Compared with the provinces, relatively few public health–related laws have been passed by the federal government in response to COVID-19. Similar to the CDC in the United States, the PHAC opted to issue guidance and make recommendations that provinces could then decide to adopt or not. Indeed, as we show later, an absence of federal policies would not have significantly reduced the overall PPI in most of Atlantic Canada.

Analyses

We conduct several exercises to compare the values of the PPI between Canada and the United States. We first explore the relationship among the state and provincial values of the PPI, federal PPI and the total PPI experienced by citizens. We next look at the sub-national units’ cumulative PPI levels throughout this period compared with an early measure of COVID-19 threat, mortality rates. Finally, we explore and compare the institutional and political origins of policies contributing to the PPI.

Federal and Sub-National Contributions

Figures 1 and 2 present the values of PPI by sub-national jurisdiction and federally for the United States and Canada, respectively, over time. The dashed lines represent the values of the sub-national indices, the solid line represents the value of the federal index, the dashed bold line represents the average value of the sub-national indices weighted by the units’ shares of the population, and the dotted-and-dashed line plots the weighted average of the total PPI in the states and provinces, respectively. The intent here is not to highlight the actions of individual states or provinces but to show the variability in levels of protective policy as well as the relative timings of public health policy responses.

Figure 1:

Comparison of Protection from State-Only and Federal-Only Protective Policy Index, United States

Figure 2:

Comparison of Protection from Province-Only and Federal-Only Protective Policy Index, Canada

Avg = average.

Source: Shvetsova et al. (2020).

Avg = average.

Source: Shvetsova et al. (2020).

Figure 1 shows the values of PPI experienced by the US population. We immediately note that by the end of the period, the state PPI value for the average American was three times higher than the federal PPI value. We also see a wide disparity in the strength of policies across states, with many states having PPIs higher than 30 by the end of the onset period and others having PPIs of only around 15.[3] Figure 2 shows the same data for Canada. We see that the absolute strength of the Canadian federal policy response is higher. Similar to the United States, there is a large amount of variation between jurisdictions in the value of the PPI by the end of the period, and we note that the early-reacting states (those whose first policies were enacted by the first week of March) reacted more strongly to the pandemic than the early-reacting provinces.

Appendix A provides the information from Figure 1 for the individual sub-national jurisdictions as well as the total PPI. In Appendix B, for each province and state, we also calculate the relative value in its total PPI at the end of the period of the federal and sub-national PPIs. This should be conceptualized as follows: for the sub-national PPI, its relative value represents what fraction of the observed total PPI would have been experienced in the event of no federal action whatsoever. For the federal PPI, its relative value represents what fraction of the observed total PPI would have been experienced in the event of no sub-national action.[4] We highlight the experiences of Nova Scotians and British Columbians here. By the end of the policy period, Nova Scotians would have had the exact same total PPI in the absence of any federal action. Meanwhile, British Columbians would have had only 69 percent of their observed PPI in the absence of federal action. For the United States, all of the relative values of sub-national PPIs are above 80 percent, with several states having a share of 1 at the end of the policy period (Alaska, Illinois, Indiana, Louisiana, Michigan, Minnesota, and South Carolina). The relatively high shares in the United States suggest that federal and sub-national efforts were not well coordinated because state and federal policies were often duplicated. In Canada, with the exception of Atlantic Canada, the shares are lower, indicating less duplication of federal policy at the provincial level.

Next, we create the cumulative PPI, which for jurisdiction j is simply the sum of the daily PPI over the entire policy period (i.e., ). The cumulative PPI combines information on how early the sub-national policies started and how stringent they were into a single measure. Both of these aspects were suggested by epidemiological research to be essential for combating the spread of the coronavirus (Hsiang et al. 2020; Pueyo 2020). Figure 3 shows the cumulative PPI for all the states and provinces. We note that the provinces are distributed roughly in line with the states, suggesting that the speed and level of sub-national response were not significantly different between countries. The cumulative PPI also provides us the opportunity to showcase the relative strength of overall policy response between the United States and Canada.

Figure 3:

Cumulative PPI Rankings: (a) US States and (b) Canadian Provinces

PPI = Protective Policy Index.

Sources: Shvetsova et al. (2020) and authors’ calculations.

To check the robustness of our conclusions to the alternative ways of combining the information about policies, we compute an alternative, item response theory (IRT)–based stringency index. Following Armstrong et al. (2020), we use the emIRT package (Imai, Lo, and Olmsted 2016) to recover a latent policy aggressiveness score for each government and day in our sample from the observed public health measures. The resulting sub-national version of the index is correlated with the sub-national PPI at 0.987, whereas the resulting overall score is correlated with the overall PPI at 0.990. Unsurprisingly, the replications of the major parts of our analysis with these alternative scores do not substantively change our conclusions. We provide a detailed comparison in Appendix E in the online supplementary materials. In addition, Appendix F offers a global comparison of the PPI with Oxford COVID-19 Government Response Tracker data (Hale et al. 2020).

Early Threat Indicator

We next show the evolution of the PPI during the COVID-19 onset phase as it relates to the threat of the pandemic as perceived by decision-makers at the outset of the policy response window. Our indicator of an early threat is the mortality rate as of 22 March 2020.[5] We chose an early date because we want to gauge the jurisdiction-specific information available to leaders at the outset of the policy response window rather than look at the COVID-19 mortality numbers as the outcome of public health policies. Medical data for 22 March on COVID-19 mortality could be conjectured to inform governments’ decisions because, given the clinical progression of COVID-19 disease, the mortality indicator could not itself be a function of policies that were adopted after 4 March, or 17 days earlier (Pueyo 2020). This indicator captures the true rates at which the disease was spreading in the population in early March. Although we cannot observe such rates directly and reliably, governors and premiers would have had more intimate knowledge and expert advice that could take into account the jurisdiction-specific testing protocols and rates of positive COVID cases. We thus use the 22 March mortality data to proxy what the premiers, prime minister, governors, and president ought to have known near the start of our policy period. We chose not to use an earlier date because, before 22 March, variation in COVID-19 mortality across US states and Canadian provinces was fairly low, which makes it hard to discern the levels of threat.

Figure 4a shows states’ and provinces’ deaths per million as of 22 March 2020. Figures 4b and 4c show the intensity of the PPI on four dates during our policy period in the United States and Canada, respectively. There seems to be no relationship between this early threat measure and the strength or speed of the policy response by sub-national units in either the United States or Canada. This non-relationship is further clarified in Figure 5, which shows the cumulative PPI (speed and strength of policy response) for the states and provinces ranked by our measure of early threat (the number of deaths per million population on 22 March).

Figure 4:

COVID-19 Mortality Rates by (a) State and Province as of 22 March 2020, (b) Comparison of State PPI, and (c) Provincial PPI over Time

Figure 5:

Cumulative PPI and Initial Threat Levels: (a) US States and (b) Canadian Provinces

COVID-19 = coronavirus disease 2019; PPI = Protective Policy Index.

Sources: Dong, Du, and Gardner (2020) and Shvetsova et al. (2020).

If early threat were predictive of an early and strong sub-national policy response, the longest bars would appear at the top of the graph (if it were perfectly predictive, the bars would be sorted perfectly in descending order). In Canada, British Columbia, Ontario, and Quebec (which had the highest initial threat and arguably the highest forecasted threat because the three largest and ultimately hardest hit cities are in these provinces) are scattered throughout the distribution of Canadian cumulative PPIs (Figure 5b). Similarly, in the United States, the initial threat does not seem to correlate with the states’ cumulative PPI (Figure 5a).

To further explore this observation, we conduct a series of Granger causality tests with mortality rates and incidence rates as the independent variable and the sub-national PPI as the dependent variable in each sub-national unit. The Granger causality test rejects mortality and incidence as an explanation for public health policy stringency. There is heterogeneity in these results but not in excess of that expected in a sample of this size (60 panels). We include the estimates from the Granger causality tests in Appendix G.[6]

Origin of Protective Policy

We conduct one final exercise with the PPI data to highlight the differences in the origin of sub-national protective policy between the United States and Canada. As states, and state executives in particular, have engaged in massive, extraordinary policy-making, their role has become greater than has been customary in the US political process in the prior period. Figure 6 shows the origin of the policies in our dataset, and Figure 6a highlights the substantial role US governors took in establishing policy compared with state legislatures and courts. At the state and province levels, Canada and the United States seem to differ in which branch of government took the lead in policy-making. In Canada, provincial legislatures continued to meet virtually or in a hybrid semi-in-person manner (Rayment and VandenBeukel 2020). Meanwhile, nearly all US state legislatures decided to close their legislative sessions, deferring policy-making leadership and responsibilities to the state governor, opting not to continue to meet and legislate. The judiciary branch in both countries continued to conduct businesses and hear cases virtually, albeit at reduced capacity (Malloy 2020; Puddister and Small 2020; Rayment and VandenBeukel 2020). For the most part, courts have allowed the other branches to create policy largely unobstructed but, with some policies pushing constitutional boundaries, will likely expand their caseload involving COVID-19 policies (Macfarlane 2020). Figure 6 does, however, show that other state government branches have acted in response to COVID-19 and will possibly expand their role in the future.

Figure 6:

Public Health COVID-19 Policies by Initiating Actor: (a) US States and (b) Canadian Provinces

The more than 150 policy items adopted by Canadian provinces during this period in more than 70 policy episodes are categorized by their institutional origin in Figure 6b. When more than one type of institutional actor joined in making the policy decision, we credit all of them as the originators of that policy. In summary, as Figure 6b shows, more than half of provincial protective public health policies had their origins with the provincial governments (premiers, single ministers, groups of ministers, or cabinet committees). More than a third of overall policy production was due to public agencies, such as health officers, or the provincial public service. Provincial legislatures joined with governments in public health policy-making on a few occasions, but COVID-19 provincial legislative output was mostly economic measures rather than protective policies. There was a lot of heterogeneity in the institutional origins of protective policies across provinces, with some dominated by political actors (e.g., François Legault in Quebec) and others relying on their chief health officers (e.g., Bonnie Henry in British Columbia).

PPI = Protective Policy Index.

Sources: Shvetsova et al. (2020) and authors’ calculations.

COVID-19 = coronavirus disease 2019.

Sources: Dong et al. (2020), Shvetsova et al. (2020), and authors’ calculations.

That the COVID-19 initial protective policy response weighs so much more heavily with public agencies in Canada than in the United States is, perhaps, the most striking contrast between the two federations. We conjecture that the blame avoidance theory and the differences in health care bureaucracies between the two countries offer a plausible explanation for this observed difference. Because efficacy of certain policies, especially amid emergencies, is ex ante unknown, political incumbents are motivated to strategically delegate decision making if possible or otherwise make decisions that would minimize subsequent blame (Hood 2010; Weaver 1986). These incentives are particularly strong when the outcome in case of failure can be catastrophic. We conjecture that the overall desire to escape blame had a two-pronged effect on the behaviour of the political incumbents during the high-uncertainty COVID-19 onset phase. First, it led national incumbents to either explicitly delegate or implicitly accept sub-national leadership in public health policy response. Second, where individual accountability has made blame avoidance not feasible, political incumbents opted for a precautionary strong initial response. The institutional design in Canada, both at the constitutional level and in health provision, enabled a substantially greater degree of delegation of protective policy-making. It enabled delegation of policy-making to the health bureaucracy, and the provincially centred health agencies were implicitly the leads in the actual policy-making, as indicated by the generally high visibility of public health chiefs both federally and across provinces.[7] In the United States, although the national government had stepped back, the constitutional separation of powers heightened governors’ individual accountability for their states’ health outcomes, and the relatively weak sub-national health bureaucracies did not play a comparably active role in policy-making.

In the United States, given the unpopularity of lockdowns for various reasons that are beyond the scope of this work, the desire for federal divestiture from policy choices acquires additional rationale (even disregarding potential constitutional reasons for so doing). This may also explain the low involvement of state legislators. In Canada, blame avoidance is achieved by relying on institutions such as the PHAC to essentially create policy. The voices and faces of unelected public health experts such as Bonnie Henry and Theresa Tam have been placed in the spotlight.

We believe that there are shared as well as divergent experiences in the policy responses of the United States and Canada. The nascent literature on Canada’s COVID-19 policy response spans the Canadian political landscape, which could result in a lot of potential for comparative work with the United States and other federations. In the discussion section, we rely on this emergent literature to attempt to explain the differences between the Canadian and US COVID-19 policy responses that we have demonstrated here.

Discussion and Conclusion

Decision-making redundancies in democratic federations allow multiple officials at different levels of government to quickly respond to crises. The responses to the COVID-19 pandemic in the United States and Canada both illustrate this point, with the sub-national governments taking the lead in adopting policies to respond to the crisis. Yet the exact mechanisms by which these policy responses were achieved in both countries varied. In Canada, public health care officials at the provincial level played a key role in developing policy, whereas in the United States governors took the lead.

Other recent scholarship has also examined how federal countries have responded to the COVID-19 crisis. Paquet and Schertzer (2020) apply the concept of complex intergovernmental problems to explain the Canadian response to the crisis, but they also offer the United States as one example of an appropriate comparison case. In addition, Beland et al. (2020) suggest that the COVID-19 crisis could be a critical juncture in Canadian politics that will lead to institutional change in Canada, affecting the equilibrium of its federal system. We reach similar conclusions for the United States and Canada here and explore the question from a system-theoretic perspective elsewhere.

Beyond just examining the federal and sub-national dimensions of policy-making during the pandemic, other scholars have examined the role that partisan and ideological considerations played in the responses. Of note, Pickup, Stecula, and Van der Linden (2020) find that although at the national level, US President Donald Trump and Canadian Prime Minister Justin Trudeau faced their first reported COVID-19 positive cases around the same time, they took different public stances in leading their respective countries. In addition, Republicans at the sub-national level in the United States as well as in the media tended to follow President Trump’s downplaying of the pandemic’s severity, whereas Democrats adopted a more quick and comprehensive policy response (Merkley et al. 2020; Motta, Stecula, and Farhart 2020).

Given that US states, and particularly governors, took the lead in responding to the crisis in the United States, it is important to take into consideration how voters evaluated these responses. At least during the onset period, it appears that residents in every state rated their state governor’s response higher than that of the president (Lazer et al. 2020). As time progresses, and as the implications of state- and federal-level policy responses become more apparent, it would be worth examining whether these attitudes persist or change.

In contrast to the United States, in Canada the public and elites appeared to be unified across party lines in supporting aggressive policy responses to the pandemic (Merkley et al. 2020). Canadians viewed measures such as mask-wearing as a means to protect others, not just themselves (van der Linden and Savoie 2020). Canadians also appeared to have formed these policy preferences on the basis of the perceived seriousness of the pandemic and trust in the ability of government to lead appropriately (Mohammed, Johnston, and van der Linden 2020; Sevi et al. 2020). As a result, policy-making has been more cooperative in the Canadian experience. In future research, it would be worth examining whether variations in trust between US and Canadian citizens in their respective governments’ ability to respond to the crisis influenced policy responses at all levels of government. Our analysis also suggests the possibility of exploring in future research whether the popular politicization of pandemic response was endogenous to the types of agent with whom pandemic public health policies have originated.

Beyond examining national, state, and provincial responses to COVID-19, Armstrong and Lucas (2020) have examined responses at the local and municipal levels in Canada and the United States. Their findings suggest that both countries saw considerable levels of policy response. They also find that in both systems, local and municipal policy was partially determined by state and provincial governments. Aggressiveness of policy response in Canadian municipalities was argued to be a factor of population size and number of cases, and ideology and geography seemed to play a modest role.

The results of our analysis and the aforementioned studies provide insight into the effects that crises can have on altering the policy-making processes in federations. In the field of constitutional political economy, the much-regarded theory of federal institutional balancing finds that the self-interested behaviour of political entrepreneurs in elections and at different levels of governments leads to self-enforcing institutional balancing, because agents continuously reach compromises and form alliances to capture momentary political gains (see, e.g., Benz and Sonnicksen 2017; Erikson and Filippov 2001; Lecours 2019; Sbragia 2002; Thorlakson 2007). Whether the past federal balance in the United States and Canada will return or whether the COVID-19 crisis upended this balance should be a topic of future research.

Click here for additional data file.

Table B.1:

US States

State	Sub-National PPI/Combined PPI	Federal PPI/Combined PPI	Unique Sub-National Policies	Unique Federal Policies	Sub-National PPI	Federal PPI	Combined PPI
AK	1.00	0.17	0.83	0.00	35	6	35
AL	0.88	0.24	0.76	0.12	22	6	25
AR	0.83	0.33	0.67	0.17	15	6	18
AZ	0.85	0.30	0.70	0.15	17	6	20
CA	0.85	0.30	0.70	0.15	17	6	20	CO	0.97	0.18	0.82	0.03	33	6	34
CT	0.88	0.24	0.76	0.12	22	6	25
DE	0.94	0.19	0.81	0.06	30	6	32
FL	0.97	0.21	0.79	0.03	28	6	29
GA	0.88	0.24	0.76	0.12	22	6	25
HI	0.97	0.19	0.81	0.03	31	6	32
IA	0.84	0.32	0.68	0.16	16	6	19
ID	0.85	0.30	0.70	0.15	17	6	20
IL	1.00	0.20	0.80	0.00	30	6	30
IN	1.00	0.18	0.82	0.00	33	6	33
KS	0.87	0.26	0.74	0.13	20	6	23
KY	0.96	0.22	0.78	0.04	26	6	27
LA	1.00	0.21	0.79	0.00	28	6	28
MA	0.87	0.26	0.74	0.13	20	6	23
MD	0.92	0.24	0.76	0.08	23	6	25
ME	0.89	0.22	0.78	0.11	24	6	27
MI	1.00	0.19	0.81	0.00	32	6	32
MN	1.00	0.20	0.80	0.00	30	6	30
MO	0.85	0.30	0.70	0.15	17	6	20
MS	0.88	0.25	0.75	0.13	21	6	24
MT	0.97	0.20	0.80	0.03	29	6	30
NC	0.83	0.33	0.67	0.17	15	6	18
ND	0.86	0.29	0.71	0.14	18	6	21
NE	0.94	0.35	0.65	0.06	16	6	17
NH	0.92	0.24	0.76	0.08	23	6	25
NJ	0.88	0.23	0.77	0.12	23	6	26
NM	0.88	0.24	0.76	0.12	22	6	25
NV	0.88	0.24	0.76	0.12	22	6	25
NY	0.88	0.23	0.77	0.12	23	6	26
OH	0.86	0.27	0.73	0.14	19	6	22
OK	0.96	0.21	0.79	0.04	27	6	28
OR	0.88	0.24	0.76	0.12	22	6	25
PA	0.86	0.27	0.73	0.14	19	6	22
RI	0.89	0.22	0.78	0.11	24	6	27
SC	1.00	0.18	0.82	0.00	33	6	33
SD	0.83	0.33	0.67	0.17	15	6	18
TN	0.87	0.26	0.74	0.13	20	6	23
TX	0.88	0.24	0.76	0.12	22	6	25
UT	0.97	0.21	0.79	0.03	28	6	29
VA	0.88	0.24	0.76	0.12	22	6	25
VT	0.97	0.19	0.81	0.03	30	6	31
WA	0.86	0.27	0.73	0.14	19	6	22
WI	1.00	0.18	0.82	0.00	33	6	33
WV	0.96	0.22	0.78	0.04	26	6	27
WY	0.86	0.29	0.71	0.14	18	6	21

Notes: Federal PPI = overlapping policies + unique federal; federal share = (federal PPI)/(federal PPI + unique sub-national policies); sub-national PPI = overlapping policies + unique sub-national; PPI = Protective Policy Index; sub-national share = (sub-national PPI)/(sub-national PPI + unique federal policies).

Source: Authors’ calculations.

Table B.2:

Canadian Provinces

Province	Sub-National PPI/Combined PPI	Federal PPI/Combined PPI	Unique Sub-National Policies	Unique Federal Policies	Sub-National PPI	Federal PPI	Combined PPI
AB	0.78	0.41	0.59	0.22	21	11	27
QC	0.76	0.32	0.68	0.24	26	11	34
BC	0.69	0.42	0.58	0.31	18	11	26
SK	0.88	0.33	0.67	0.12	29	11	33
ON	0.81	0.34	0.66	0.19	26	11	32
NL	0.93	0.39	0.61	0.07	26	11	28
NS	1.00	0.31	0.69	0.00	35	11	35
PE	0.93	0.37	0.63	0.07	28	11	30
MB	0.86	0.38	0.62	0.14	25	11	29
NB	1.00	0.35	0.65	0.00	31	11	31

Notes: Federal PPI = overlapping policies + unique federal; federal share = (federal PPI)/(federal PPI + unique sub-national policies); sub-national PPI = overlapping policies + unique sub-national; PPI = Protective Policy Index; sub-national share = (sub-national PPI)/(sub-national PPI + unique federal policies).

Source: Authors’ calculations.