Institutions and the politics of agency in COVID-19 response: Federalism, executive power, and public health policy in Brazil, India, and the U.S.

Abstract The COVID-19 pandemic of 2020 was one of the rare events that shocked almost every world government simultaneously, thus creating an unusual opportunity to understand how political institutions shape policy decisions. There have been many analyses of what governments did. We focus instead on what they could do, focusing on the institutional politics of agency – how institutions empower rather than how they constrain, and how they affect public policy decisions. We examine public health measures in the first wave (March-September 2020) in Brazil, India, and the U.S. to understand how the interplay of institutions in a complex federal context shaped COVID-19 policy-responses. We find similar patterns of concentrated federal executive agency with limited constraints. In each case, when federal leadership failed public health policy responses, federated, subnational states were left to compensate for these inefficiencies without necessary resources.


Introduction
Donald Trump of the U.S., Jair Bolsonaro of Brazil, and Narendra Modi of India were three of 's more controversial leaders.While few countries controlled COVID-, their countries are among the ones that have suffered most in the COVID- pandemic.All three downplayed the pandemic, left key elements of response to federal states, removed public health restrictions far earlier than most public health experts counselled, showed negligible interest in the impact of the pandemic on economic or health inequalities, and were rewarded with large surges of cases and deaths.Bolsonaro and Trump not only caught COVID- themselves, they also supported quacks and quack remedies.
It is natural to look to the content of the three men's ideology and politicsto their populism, nationalism, authoritarianism and hypermasculine political stylefor an explanation of their failures (Waylen, ).It is interesting that such leaders would perform so badly in the face of a cross-border pandemic that required behavioural changes justified by scientific advice, even if the evidence is still unclear as to populism's effects (Falkenbach and Greer, ; Kavakli ).But focusing on their agency is only one side of the coin.The other is to ask what agency they had.Our question is: why were these three leaders able to produce such similar and destructive results despite vast differences within and between their countries?Our goal is to identify explanatory similarities in otherwise very different places, by focusing on the institutional structure of agency among the different leaders.We argue for a focus on the institutional politics of agency, asking why countries did what they did by asking who could decide what they would do.These three leaders simply mattered more to their countries' fates than the leaders of countries with different institutions.
We focus on who acted rather than the content or health effects of their decisions.We can stand on the shoulders of generations of institutionalist scholars of social policy in doing this (Lynch and Rhodes ), even if "disease outbreaks and political institutions have been under-studied in comparative politics" (Kavanagh and Singh ).Institutions shaped the agency of governments in both public health, where in  the main issue was adoption of nonpharmaceutical interventions (NPIs) such as school closures (Markel et al., ), and in social policy to compensate for the costs of the pandemic and NPIs such as unemployment insurance or blanket payments (Greer et al., ).Following Lijphart () and others, we focus on the explanatory power of two key political institutions that shape agency: federalism and the nature of executive powers.We then present the rationale for our case selection and methods, and subsequently analyse the institutional contexts and the COVID- responses of governments in our cases.

Institutions and agency
Many of the classic works of institutionalism are about constraints on political actorson veto points (Immergut, ) and veto players (Tsebelis, ), on how rules enable games by making activity predictable (North, ), on how path-dependent institutions constrain policy change (Pierson, ), or on the ways that demos-constraining federations (Stepan, ), counter-majoritarian        -19   courts (Lemieux and Watkins, ), and other such institutions constrain agency.Institutions constrain rather than empower, and even when an actor is empowered, in these theories, it is often through a transactional mechanism, veto power or agenda control, rather than the simple, hierarchical ability to act (Taylor et al., ).But institutions do enable action.They create and redistribute agency, vetoes and transactions.Here, we explicitly consider this intertwined relationship between agency and political institutions, where dependent upon the structure of political institutions, there may be fewer constraints on the power of individual actors.In our case, that means fewer checks on power of the executivewith the consequences of reduced accountability dependent upon the benign nature of the executive in their ideological context.The influence of ideology and the changing nature of ideology in our federations of interest will be discussed shortly.Overall, we expand on existing literatures to consider not only the influence of structure on enabling agency (power to act), but on the production of policy outcomes from created agency (Pierson ).Whereas much literature has focused on degrees of discretion, we consider the consequences of expanded discretion for policy.
Pandemic response involves, almost by definition, agency, and gives us a chance to expand beyond the often, slow-moving processes such as life expectancy that have dominated research on the impact of institutions on health (Kavanagh and Singh ), or economic issues that have dominated the comparative study of government responses to shocks.The story of COVID- response is substantially about how institutions gave some people critical agency in the pandemic's hectic early months and what they did with it.

Federalism
A federation is a political system in which there are multiple territorial units of general-purpose government whose independence is constitutionally guaranteed (Elazar :).The impact of federalism on public policy and welfare is much disputed.Some argue that it produces adaptiveness, innovation, and the optimal size of government (e.g., Weingast , Tarr ).Others argue it leads to rigidity, less responsive policy, and less generous welfare states (e.g., Rueschemeyer et al., ).Some scholars contend it does not matter, as did Riker (see Stepan, ) or matter in a consistent way across contexts (Erk, ).
In principle, federalism means leaders of federal and state governments have agency to take and make decisions (Bednar ).In practice, state governments' autonomy varies with institutional design.State autonomy is systematically less than that of the federal government, while existing within a framework created by federal policies and federal party politicspolicy legacies and constituencies so deeply entrenched as to constitute a "policy state."States nevertheless diverge, albeit within the constraints set by financing, law, and politics of central states.The clearest effect of federalism on social policy is simply variation rather than the system-wide upwards or downwards effects on generosity that so many thinkers sought to find (Kleider, ; Greer, ).A policy area left to federal states will be one in which policies vary, even in areas where federal levers such as judicial oversight or conditional grants can shape what states do.

The Executive
If we are examining the politics of agency, the executive is likely to be crucial.Executives might especially matter in federations, where any centralized policy responses require national oversight that may come from or be swayed by the executive.We start with the foundational distinction between a parliamentary and a presidentialist system.A presidentialist system has, in Linz's definition: "an executive with considerable constitutional powers-generally including full control of the composition of the cabinet and administration : : : [and that] is independent of parliamentary votes of confidence." Linz continued that "two things about presidential government stand out.The first is the president's strong claim to democratic, even plebiscitarian, legitimacy; the second is [the president's] fixed term in office" (Linz, : -).In other words, the president's accountability is directly to the electorate, not to the other branches of government, and is on fixed schedules.Separate electoral mandates enable gridlock if the legislature and president are not of the same party and can induce the president to test the limits of the position's power.Unlike in parliamentary systems, the president can claim a personal national mandate, need not be formally accountable to a party, and can sometimes strike out against their own party or legislature.
Subsequent scholarship qualified the impact of presidentialist political systems.In part, the details of different presidential arrangements (e.g., term limits) mattered, and semi-presidential systems in which a president co-exists with a prime minister accountable to the legislature turned out to have their own distinctive dynamics (Elgie, ).More broadly, most political institutions' effects can be best understood by looking at their interaction with other political institutions and forces such as electoral rules and party systems (Taylor et al., ).Linz' insight has been substantially subsumed into broader theories of institutions (Elgie, )even if the basic intuition, that presidentialism creates distinctive forms of instabilityhas survived.
A parliamentary government can under some conditions be as forceful and difficult to hold accountable as a president.That is why Lijphart () argued that presidentialist and Westminster parliamentary systems could be clustered as majoritarian governments, where decisiveness was structurally privileged over consensus (and agency deliberately created and focused in one place).
       -19   In a Westminster-style majoritarian system with a parliamentary government and first-past-the-post electoral system, a Prime Minister can enjoy a great deal of agency to act, or to not act, without winning a majority of the popular voteas we see in our case of India.

Cases
Exogenous shocks like COVID- represent an opportunity to explore the effects of institutions, in the same way economic shocks have been a longstanding source of cases for the comparative study of institutions.An infectious disease that spreads rapidly, does not respect borders, and requires countries to take similar precautions is ideal to see how political institutions matter.We focus on the period of January to September , covering what many have labelled the "first wave" of the pandemic (Baldwin, ).During this time, public health responses relied on non-pharmaceutical interventions (NPIs) to mitigate viral spread.Vaccination campaigns had yet to start in most places in the world, and there was great uncertainty about when vaccine clinical trials would be completed, and vaccines would be authorized for use.Policy was contingent and uncertain, therefore, an ideal laboratory to explore agency.In periods of crises, agency matters.
We define governments' response to Covid- as public health policy.In , public health policy primarily meant NPIs and the construction of testtrace-isolate-support (TTIS) systems.NPIs are public health actions to slow or stop the spread of disease that do not involve vaccines or medicines (Markel et al., ).In  prominent NPIs included hand washing, social distancing, travel restrictions, school closures, restrictions on businesses or closures of activity in places or sectors ("lockdowns"), mask mandates, and restrictions on working and socializing.NPIs could be extremely broad, effectively closing entire cities, sectors, or countries, or could be relatively limited.TTIS systems are a package of interventions (Jarman, ): testing extensively (to find people with virus), contract tracing (to find out whom they might have infected and from whom they caught it), isolation (keeping infected people away from the public), and support (ensuring that infected people have what they need to isolate, e.g., income, food, support for dependents while their caregivers were in isolation, a secure place to live, health insurance).Together, NPIs and TTIS systems were the core of the successful efforts to manage the pandemic.The success of these measures required enormous economic investment from governments, fostering close collaborations between central banks and fiscal authorities rarely seen outside of the  financial crisis (Tooze, ).Through government bond buying and direct support to firms, among others, the US Fed, the Brazilian Central Bank, and the Reserve Bank of India, set the groundwork for effective use of NPIs and TTIS.While these were the most effective measures and thus our definition of successful pandemic-approaches, adoption and enforcement varied widely, which we examine in the context of executive influence.
We use a qualitative comparative approach for several reasons (Mahoney and Thelen, ).First, it is challenging to compare country outcomes given overdetermination (more possible causes than cases), the idiosyncrasies of every country, and the stickiness of institutional variables.Second, institutions matter in configurations (Greer et al., ).Interactions between variables are ubiquitous, and easier to identify in case studies.Third, tracing processes allows us to identify the use of agency, including decisions to act or not and who made them.
Our cases are Brazil, India, and the U.S., which are all federal and allow us to observe the operation of federalism, while offering different models of legislative and executive politics.The U.S. has two polarized but organizationally weak parties and a directly elected executive; Brazil has a directly elected executive and famously weak parties, and India a parliamentary system at the federal and state levels with many parties.The U.S. is very rich; Brazil and India and to a lesser extent the U.S. have strikingly high levels of economic inequality; their cultures, histories, and party politics are obviously quite different, and their federal systems operate in significantly different ways.If we can identify similarities in the dynamics of agency in the crisis, we may add to the fund of knowledge about conditions under which a given factorhere, institutional formmay matter in explaining policy.
These three leaders have been termed right-wing populist (Meyer, ).They claim to speak for ordinary citizens rather than 'elites' and emphasize family values and individual and national autonomy in support of reduced government regulation.Notably, Populist leaders set their own definition of who they consider to be an ordinary citizen or 'the people' often taking nationalist or ethnocentric sentiment, generating animosity towards 'outgroup' populations to engender support for their platform (Howell and Moe, ).Populist leaders in Brazil and the US sought to diminish the potential threat posed by the pandemic, while Modi used the crisis to further the cultural conflicts that brought him to power (Meyer, ).Instead of looking to the three leaders' ideology and politics, we argue that their ability to shape the national political response was rooted in the political power accorded them by their country's constitution.

Brazil Institutions
The Brazilian presidency is one of the most powerful in the world (Shugart and Carrey, ).Brazilian presidents are endowed with strong constitutional powers, including the prerogative of issuing presidential decrees which can create, modify, or regulate social programs; extensive reactive power, that is, presidents can partially or totally veto bills passed by the Congress; and, particularly        -19   important for public policies, freedom to form cabinets without formal approval of Congress and exclusive initiative over budgetary matters (Figueiredo and Limongi, ).
Brazil's political system has a record number of political parties.This fragmentation complicates executive government rule.When presidents allocate positions in the vast federal administrative empire, they usually respect the share of Congressional seats of their party coalition members (Amorim Neto, ).Without sharing executive power with their allies, it would be difficult for Brazilian presidents to secure a legislative majority and govern (a concept known as "coalitional presidentialism").For this reason, the bargaining space between coalition members and the executives in cabinet formation (including the Minister of Health, MoH) comes from the particular shape of the party system.Finally, as in other countries, the chief executive also holds non-legislative prerogatives allowing them great visibility for speaking directly to voters through radio and televised speeches (Amorim Neto, ).This is a powerful agenda-setting instrument.The country's electoral rules offer strong incentives for politicians to behave individually, developing direct links with constituency groups rather than mediating relations through political parties (Alston et al., ).Presidents can use addresses not just to send public interest messages but also to put forward personal, particularistic goals.
Brazilian subnational governmentsfederal statesare a prominent check on presidential power.Coordination among these levels of government is a major challenge in policy and administration.For instance, the Ministry of Health has the constitutional mandate of coordinating Brazil's extensive public health system, including  states and more than , municipalities; both levels are elected and responsible for healthcare provision.Their activities are substantially financed through federal conditional grants, e.g., for health, as well as revenue sharing and their own taxes.
Bolsonaro is a far-right populist president who came into power in  through an alliance between the economic liberals and social conservatives.As a low-rank congressman, he never had great political aspirations but was seen as an alternative to the Workers Party's candidate (PT) after corruption allegations resulted in the impeachment of President Dilma Rousseff (Evans, ).Particularly notable about Bolsonaro's government is its decision to not follow the rules of coalitional presidentialism and attack the Supreme Court at an unprecedented level (Amorim Neto and Alves Pimenta, ).On the one hand, he never commanded a stable majority in Congress; on the other hand, he has had great freedom to control the bureaucracy and public policies (including health).Bolsonaro's response to COVID- reflects his ongoing prioritization of business interests, and like the American populist president, Donald Trump, was keen not to 'stop' the national economy.Also like Trump, he gained notoriety for supporting protests against government lockdowns, touting unproven medicines, downplaying the seriousness of the virus, and even vocally opposing state governors' decisions to impose social distance measures.

Pandemic response
Brazil has been one of the countries most affected by COVID- (Dong, Du, and Gardner, ).Although the country was relatively well-prepared for public health crises (it had performed in an exemplary manner in previous epidemics and has one of the largest public health infrastructures in the world), Brazil's response to the virus has been widely seen as inept (Lancet, ).Much of this can be attributed to President Bolsonaro's strong scepticism toward the science of COVID-; reinforced after returning from a visit to the U.S. where he met with President Trump.
At the outset of the pandemic, the MoH was Henrique Mandetta, a physician and politician from an important party allied to the president.Bolsonaro's pressure on the MoH has no precedent in Brazil.In April, when Brazil' COVID- cases were second only to the U.S., Mandetta was fired for threatening the President's political dominance and his pseudo-scientific rhetoric.The President then appointed a respected physician, Nelson Teich.However, due to his vehement disagreement with President Bolsonaro's plans to adjust clinical protocols for COVID- treatment, Teich resigned less than a month after taking the position.Teich was replaced by active-duty Army General Eduardo Pazuello, who yielded to the adjustment of the clinical protocols, a decision that was highly criticized by the public health community (Abrasco, ).Pazuello also reformulated the disclosure of epidemiological data, announcing only information about death and confirmed cases in the previous  hours rather than accumulated deaths and infections (Machado and Fernandes, ).In response, a consortium of media organizations established an online database updated daily to monitor and compare the official data provided by the Ministry of Health (G, ).Although Pazuello's decision was subsequently repealed by the Supreme Court, data quality was severely compromised (Idrovo et al., ).
The pandemic struck the country during an economic crisis, demanding increased social expenditure, jeopardizing Bolsonaro and his Ministry of Economy's austerity policies.The executive government's inaction delayed the much-needed social policies to support social distancing measures (Fonseca et al., ).Congress had to strongly pressure the Ministry of Economy for a cash transfer program (Piovesan and Siqueira, ).Bolsonaro vetoed several parts of COVID- related-legislations issued by Congress (e.g., mandatory use of masks in religious sites) (Bertoni, ).
Thanks to the authority of state governments over health policy, Brazil was able to secure some level of social distancing, NPIs, and coordination with the WHO measures (Fonseca et al., ).For almost three decades, state governments have had limited influence on Brazilian national politics given the institutional powers of        -19   the executive and the way that tax resources are distributed, but in the vacuum of federal NPI leadership they enacted measures and communicated public health information.The President tried to question subnational governments' authority over pandemic management, but the Supreme Court ruled in favour of governors.Bolsonaro then used his decree power to list essential business that should remain open (e.g., beauty salons), arguing subnational governments went too far in social distancing measures, damaging to the economy (Reuters, ).Most state governments remained firm in their support of social distancing, business closures, and warnings against therapies that had yet to be tested.Petherick et al. () demonstrated great variation in the severity of social distancing measures supported by federal and subnational governments, with the latter contributing more to Brazil's country-level stringency scores.
Bolsonaro took advantage of the division of authority over the COVID- epidemic to adopt a blame-avoidance strategy.The President used his power to address the nation on national TV seven times between March and September , attacking subnational governments, sowing doubts about the seriousness of the pandemic, and fostering public demonstrations against social distancing. Polling suggests Bolsonaro was able to pass the blame onto state governors, whilst claiming credit for social policies (particularly cash transfers).Whether this was a coincidence, or a savvy political strategy is unclear.

United States Institutions
The U.S. is a presidentialist, federal system.American citizens elect representatives to make policies on their behalf, at the state and federal levels.States are responsible for policy design and delivery across many types of social and health policy but rely substantially on the federal government to fund these programs.While federal legislation and regulation comes with standards states must carry out or adhere to in order to receive funding, states are often seen as 'laboratories of democracy', where they may experiment in the policies they design and deliver, leading to widely heterogenous policy approaches across the nation (Tarr, ).
At the federal level, a president is elected to four-year terms.The discretion given to the presidency has increased over time, most notably over the past two decades.Growth in executive power has tracked with increased polarization and partisanship, leading to stagnation and dysfunction in congressional policymaking, providing more opportunities for the executive branch to act without being checked by the legislature (Whittington and Carpenter, ).A primary reason is the combination of strong partisanship with organizationally weak parties (Azari, ).While political parties are an organizing force (Aldrich, ) influencing policy agenda setting and adoption (Bawn et al., ), increasing partisan polarization, growing fragmentation within the Republican Party (Williamson et al., ), and the increasing use of undemocratic tactics by Republicans made Republican party politics in particular more susceptible to sway by factions, or individual political actors at the head of their party, like President Donald Trump (Howell and Moe, ).As Drezner () wrote, "political architects in both major parties had worked at building the presidency into the most powerful position in the world.As polarization gripped Congress, the president was viewed as the last adult in the room.And then someone with the emotional maturity of a small child was elected to that office."

Pandemic response
Despite high levels of public health capacity and estimated pandemic preparedness, the U.S. would have always faced challenges and risks during a pandemic (Singer et al., ) resulting from: pre-existing health disparities arising from systemic racism; underinvestment in social spending (Hacker, ); fragmentation and decentralization in public health and health care systems (Greer and Singer, ).U.S. disaster responses rely on federal action, specifically executive action that subsequently delegates responsibility to federal agencies to coordinate national strategies and distribute resources.Yet instead, during the COVID- pandemic, limited federal action, coordination failures and executive politicization of COVID- unprecedently placed the onus on states to independently respond and coordinate responses to COVID-, without sufficient capacity.
Federal disaster responses, especially during public health crises like disease outbreaks, require executive action to declare an emergency, coordinate overall approaches across the vast and fissiparous federal bureaucracy, and delegate responsibility to specialistswhich in the US is the Centers for Disease Control and Prevention (CDC), and previously included the National Security Council Directorate for Global Health Security and Bio-defence, which was disbanded in  (Riechmann, ).Trump side-lined scientific experts in the federal public health bureaucracy (such as Anthony Fauci), and promoted scientific misinformation about COVID- and its prevention and treatment (Cancryn et al., ).The executive branch made active efforts to constrain and limit the influence and expertise of the federal public health bureaucracy, including having politically appointed officials pre-screen CDC pandemic response materials (Diamond, ).
States struggled to make pandemic policy choices by themselves, lacking federal guidance and without federal resources and funding in many cases (Rocco et al., ).The federal government politicized aid distribution: the executive branch withholding aid to states of the opposite political party or in opposition to the administration.States held the power to enact NPIs and used them quite differently.In some states, state courts or legislatures acted,        -19   along partisan lines, to constrain public health measures (Quinton, ).Subsequently, in Michigan or Wisconsin even the state government could not use powers constitutionally reserved to states.Finally, the pre-existing risks of health disparities in the U.S. were exacerbated by limited federal coordination and constrained federal resources, with disproportionate mortality rates among racial or ethnic minority groups and little attention to describing these disparities let alone ameliorating them (Singer et al., ).
The response to COVID- involved not just a public health response, but also social policy, which is inextricably related to public health.As discussed, the weak, fragmented, and broadly market-based safety net system in the U.S. would have always posed challenges to Americans during a pandemic.Responding to economic downturn, the federal government passed the Coronavirus Aid, Relief, and Economic Security (CARES) Act, signed into law on March , .Compared to the public health response, CARES was initially very successful, and notably boosted income by nearly doubling unemployment benefits and sending stimulus checks directly to Americans (Ganong et al., ).
Despite this early success, lacking the federal public health action necessary to drive down COVID- rates positioned CARES for failure, by prolonging socioeconomic distress arising from unrestricted disease spread.Subnational governments were unable to compensate for federal failures on their own by virtue of balanced budget requirements in state constitutions, and municipal government's attempts to avoid bankruptcy.Thus in autumn, state NPIs were usually weaker than they had been in spring, even as cases, hospitalizations, and deaths hit levels not seen in spring.

India Institutions
India is a parliamentary federal democracy (Heitzman and Worden, ) with a UK-style Westminster government system with first-past-the-post elections to territorial districts at the federal lower house (Lok Sabha) and state levels.India has eight national parties,  state parties, and  other parties competing to represent . billion people (Sangita, ).Narendra Modi of the Bharatiya Janata Party (BJP) became the th Prime Minister of India in  following his tenure as Chief Minister of Gujarat.He was re-elected in .Modi and the BJP married anticorruption policies with the argument that he would be a "watchman" for the country's wellbeing (Adeney, ).Reflecting the BJP's deep roots in Hindu religious mobilization, it appealed to Hindu majority states (Biswas, ).
Outside "union territories" controlled by the central government, states have their own legislatures.These state legislatures make laws regarding criminal justice, education, public order, lands, forests, and substantial areas of taxation, and depend substantially on their own taxes and debt, meaning the economic contraction of COVID- left them "grappling with the pandemic with constrained fiscal space" (RBI, :).Healthcare is constitutionally a state-level responsibility but the central government shares in state infrastructure and resource funding.States have authority to manage epidemics and disasters and can declare state-level emergencies but can be overridden by a federal declaration of a state of emergency.
Resulting from this division of power, particularly in instances of healthrelated crises, the central government can either implement a unified response (such as a lockdown), or, leave authority to statesas Modi chose to do for much of .First, India's  states may vary in their approaches to health and in their management of competing priorities.Second, while central government funding to states is allocated based on various population statistics and needs, this formula might not reflect particular needs during unprecedented times such as the COVID- pandemic.State governments can, at the discretion of the central government, be left with problems they lack the resources to solve and were not prepared to address.

Pandemic response
India's pandemic response has stood out: first as an outlier in its strict initial response, and second, in the rapid surge in cases when pandemic management shifted from central authority to state-level policies.The first confirmed case of COVID- was in Kerala on January , .In March travellers returning from Italy were also deemed positive for COVID-, marking the beginning of a surge (WHO, ).
On March , the "Janata (People's) Curfew" began as the largest lockdown in India's historyfourteen hours per day lockdown.Two days later PM Modi announced a complete national lockdown including a ban on international and domestic travel, with the exception of essential workers (Press Trust of India, ).Some states (Kerala, Karnataka, and Odisha) had already previously implemented their own stay-at-home orders.Reports of the lockdown emphasized unified and standardized implementation across all states with surprising compliance, which was likely facilitated by severe penalties, mostly fines.Anticipating issues with hospital capacity, India's trainsotherwise suspended for travelwere converted into field hospitals, as were schools and colleges.Simultaneously, Finance Minister Nirmala Sitharaman announced a stimulus package for the equivalent of  billion USD, of which about  billion USD were allocated for direct cash transfers to individuals (The Economic Times, a).
This looked like as dramatic and effective a combination of NPIs and social policies as one could expect.Yet, this was the point at which purportedly        -19   promising, centralized decisions went awry (Raj, ).Over  million migrant workers were stranded in their state of work, unable to return to their homes due to the Indian Railways suspension (Athrady, ; Maji, Sushma, and Choudhari, ).The central government's stimulus package did not reach migrant workers in their region of work despite their One Nation One Ration identification card due to misinformation about eligibility, and because the process for presenting one's card and identifying oneself requires biometric measurements that would have exacerbated the risk of the infection spreading (Jebaraj, ).At this point, the central government was at a crossroads and decided to resume inter-state movement for migrant workers via railways (Mitra, ).
With migrant workers subsequently having to quarantine, and other nonessential workers staying home, the economic ramifications of a lockdown overwhelmed the central government.The national lockdown was expected to lift in April.Several states (Odisha, Punjab, Maharashtra, Karnataka, West Bengal, and Telangana) extended their lockdowns until the end of April, which actually motivated the central government to follow suit and extend the lockdown until May (Hindustan Times, ; The Economic Times, b).When the central government ultimately lifted the lockdown order in June, designated "hotspots" were exceptions to the national "Unlock" but eleven states continued their own lockdowns into July.
Positive cases and deaths continued to rise in India.Central government attempts to contain the virus were initially ground-breaking, but management of COVID- was ultimately left to the states in terms of health and health care response as well as the social policies to enable people to follow state guidelines.A crucial example was limited coordination mechanisms guiding learning continuity when schools closed in March  at which point over  million children and adolescents were out of school (Van Cappelle et al., ).As in Brazil, states created their own remote learning strategies and response plans.States created plans based on estimates of technology availability in homes; for some, YouTube and TV channels were used for telecasting lessons in the language of instruction for each school.And yet, lack of standardization and quality consistency across these different modalities and languages limited learning effectiveness.In both countries, the consequences of long school closures have been substantial.In India, limited federal involvement may have been a particular reason for inconsistencies in access and quality of materials, and thus, in outcomes (UNESCO, UNICEF, and World Bank, ).Despite the central government's ultimate resignation of power to the states, recent state elections in Bihar the country's second largest stateand  elections in Uttar Pradesh, suggest that Modi's BJP is paying little price (The Economist, ).

Conclusion: Institutions and the politics of agency in emergencies
In the countries we examined, agency distribution empowered executives: presidents (in Brazil and the US) and a powerful Prime Minister in a majoritarian parliamentary system (India).In each case federal actionand inactionwere in the gift of that leader, with hierarchy dominating what might look like transactional relationships.In all three cases, by late summer  the federal executive opted for inaction and, in Brazil and the U.S., sabotage of its own policies.In each case, therefore, state governments confronted the challenges, though with a great deal of variability in their choices, timing, resources, and effectiveness (as might be expected in diverse federal systems).In all cases, furthermore, as the pandemic dragged on it became clear that states lacked the fiscal capacity, border control, or political legitimacy to enact serious NPIs or other major public health policies without central government support.
Federalism incentivized state-level governments to respond with a variety of measures, redistributing agency, producing some insulation against federal policy failures.Yet, in each case, that contribution was limited.State agency was constrained by the real and constitutive power of the central government.In particular, states lacked fiscal resources necessary for effective large-scale public health, social policy, or economic interventions.Sustaining business closures, for example, was a challenge for state governments lacking the resources to help businesses and the temporarily unemployed, and state-level travel restrictions were difficult to maintain in integrated economies and substantially nationalized political systems.In each country, federal leadership and resources emerged as being critical to effective state responsesfederal executive agency mattered greatly even in the context of federalism.
While we focus on the first wave, we believe subsequent developments support our argument.In public health, all three federal governments supported reopening early and failed to take timely action in response to later waves, leaving agency to state governments unable to sustainably impose major public health measures or compensate in social policy for losses they entailed.Once vaccines became available, federal executives dominated decision-making on key issues, above all vaccine choices, purchasing, and decisions about imports and exports, with states relegated to vaccination strategies and occasional efforts to acquire their own supplies.
Presidentialist systems create a "single point of failure" in the executive (Fallows, ).While in principle India's parliamentary system is more flexible if perceived as an electoral liability Modi's party could have replaced himin practice the politics of the BJP meant Modi's policy choices and leadership were not effectively challenged.By comparison, the U.S., also saw a change of government in January .The Biden administration's very different approach to public health, including its concern for equity and vaccinations, showed just        -19   how much presidents matter to policy, for better or worse.In other words, we saw exactly the lack of resilience Linz would expect of presidentialist systems, and the kind of policy underperformance that Lijphart would expect of majoritarian parliamentary or presidentialist systems with limited ability to check rogue party leaders.
The main implication is that political institutions, in shaping agency, can create risks not just to democracy but to public policies.Institutions that delegate public power to a strong executive run the risk of abuse of power.Linz's initial insight about the perils of presidentialism seems substantially validated, however much of the dynamics he noted are qualified.Federalism, meanwhile, is able to supplement federal leadership partially and unevenly, but unable to provide nationwide leadership or response due to legal and fiscal constraints.While Americans, Brazilians, and Indians were safer in some states than others, none of them were as safe from the pandemic or its economic effects as they might be had they not been left dependent on state governments that were never designed to manage this kind of shock.Furthermore, interactions between and the institutionalization of hyper-partisan politics and presidentialism in the U.S. demonstrate the contextual importance of understanding potentially buffering effects of individual states, or federated powers, on an overpowered, authoritarian executive with strong party-ties.Though the Biden Administration started using executive power for a more scientifically informed, equitable public health approach, conservative state governments started to hamstring federal actions by institutionalizing policies (or the absence of policies) left over from the Trump Administration.Further, empowered Republican state executives launched lawsuits for partisan gain against Biden's federal attempts to coordinate policy levers across subnational units.
Arguments about appropriate political institutions and distribution of power are as old as political thought, as are arguments about whether political institutions really explain outcomes.Institutions never really offer a clean slate to institutional designers.But learning lessons still has value.Institutions are reconstructed and maintained every day, through actions large and small, such as the willingness of American "political architects of both parties," in Drezner's () phrase, to endlessly increase presidential agency.Future architects might wish to learn from the effects of agency in our three executive-centred federations.In the meantime, focusing on political agency in our analysis of institutions makes it clear where citizens, scholars, and the future should focus blame.There will surely be plenty of blame to go around.