Federalism and Democratic Backsliding in Comparative Perspective

Abstract

As a wave of backsliding has swept across both new and established democracies, scholars have sought to identify formal and informal institutions that can act as guardrails of democracy. But while informal norms, party structures, and formal institutions such as separation of powers have all been singled out as potential bulwarks against democratic backsliding, the literature has had less to say about what role, if any, federalism might play in stopping democracies from sliding into autocracy. While some constitutional designers and scholars have argued that federalism can help to prevent the emergence of a national tyranny, most contemporary research has emphasized the damaging effects federalism can have on democracy. In this article, we assess the relationship between federalism and threats posed by national rulers, with quantitative analysis of that relationship in countries around the world and with structured, focused comparative case studies in the United States, Brazil, Venezuela, and India. Our quantitative analysis finds no systematic relationship between federalism and backsliding, while our comparative case studies support our argument that federalism is only likely to serve as a bulwark against autocratic threats posed by national rulers under a limited set of conditions.

In The Federalist No. 51, James Madison argued that separation of powers and federalism would work together to provide “a double security” guarding American society against oppression by its rulers. Not only would legislative, executive, and judicial branches of government act as checks on one another, but the different levels of government—federal and state—would also control one another. Madison and the American Founders were by no means the only constitutional designers who believed federalism could help to prevent a democratic republic from backsliding into tyranny. For instance, the architects of the postwar Federal Republic of Germany believed that entrenching a federal division of powers was crucial to ensure Germany could not fall back into dictatorship (Currie 1994, 33–34). ¹

A great deal of contemporary research has provided an alternative perspective, highlighting the damaging effects that federalism can have on democracy and fundamental rights. This literature shows how the dynamics of federalism can foster autocratic enclaves at the state level (Gervasoni 2010; Gibson 2012; Giraudy 2015; Mickey 2015), and even undermine democracy at the federal level (Bateman, Katznelson, and Lapinski 2018; Grumbach 2022). In the US, infamously, antimajoritarian features of the federal system have facilitated racial authoritarianism and the systematic violation of African Americans’ civil and political rights. As the eminent scholar of federalism William Riker (1964, 155) once put it, “[I]f in the United States one disapproves of racism, one should disapprove of federalism.”

While threats from local authoritarian enclaves persist, since the early 2000s many democracies around the world have faced new threats from the election of national leaders who seek to use their office to roll back democracy and install themselves as autocrats. This is precisely the kind of threat that many proponents of federalism feared. They hoped that the emergence of tyranny might be checked by the fragmentation of power between the center and the states and by strongholds of democracy at the subnational level. And yet, although recent comparative work on backsliding has singled out informal norms, party structures, and separation of powers as guardrails of democracy, very little has been written about the potentially positive impact of federalism outside the United States. ² To the best of our knowledge, the literature offers no systematic, comparative analysis of what role, if any, federalism might play in protecting democracies from threats by elected autocrats.

In this paper, we aim to provide that comparative analysis. We develop a theoretical argument about the conditions under which federalism is likely to impede efforts to dismantle democracy from above, and we assess it with a mixed-methods approach, combining quantitative analysis and structured, focused, comparative case studies. We argue that federalism can discourage this form of democratic backsliding as Madison and others have anticipated, but only when two highly demanding conditions are met. First, states must have sufficient authority to present a meaningful obstacle to an aspiring autocrat who is trying to consolidate power at the national level. State governments can only exert meaningful checks on federal authority when they possess substantial control over key governmental functions, including judicial authority, administration of elections, law enforcement, and the delivery of public services. Second, who controls state governments is crucial. We should only expect to see state powers used effectively to push back against abuses of national authority when a substantial number of state governments remain under the control of opposition political parties or factions of the ruling party that are independent of the aspiring autocrat. Conversely, federalism becomes a far less effective bulwark as the ruling party expands its power across the states, and as copartisans within the ruling party begin to accept autocratic abuses.

This paper examines these issues in three sections. We begin with a theoretical elaboration of federalism’s potential effects on democratic backsliding. We acknowledge that federalism may have contradictory effects—encouraging backsliding in some circumstances and discouraging it in others. Only under the particular set of conditions mentioned above will federalism reliably act as a guardrail for democracy; we would not, therefore, expect that federalism itself would have a significant effect across the universe of cases. Next, in a quantitative analysis of all democracies between 1974 and 2021, we show that, consistent with these expectations, there is no statistically significant relationship between federalism and the occurrence, the pace, or the severity of democratic backsliding.

But the lack of association between federalism and democratic backsliding does not settle the matter. We must be cautious and explore the issue further for at least two reasons. First, the absence of any statistically significant correlation in our dataset may simply result from the fact that the universe of cases is so small. Second, as Bednar (2021, 5) notes, “Federalism has too many qualitative variations to be easily plugged in as a dummy variable to make a straightforward prediction.” Testing hypotheses based on the stark distinction between federal and unitary systems may obscure more than it reveals, for instance by ignoring variations in the degree of centralization between federal polities.

To better understand the impact of federalism on democracies experiencing backsliding—and to assess our arguments about the conditions under which federalism might act as an effective guardrail for democracy—we turn to structured, focused, comparative case studies of the only four federal democracies in the world that, according to our criteria, experienced backsliding in the period covered by our study—the United States, Brazil, Venezuela, and India. Because we want to study the impact of federalism as a guardrail in cases where democracy is challenged by an aspiring national autocrat, it makes most sense to focus on cases that have had their federal institutions tested by an episode of such backsliding, rather than on federal systems that have not faced such threats. Moreover, focusing on the US, Brazil, Venezuela, and India allows us to explore the extent to which differences in states’ institutional resources and partisan independence affect the likelihood that they can act as guardrails of democracy. In the United States and Brazil, federal systems comprised of states with substantial institutional and political independence appeared to be important factors in heading off—for now at least—the autocratic threats of Donald Trump and Jair Bolsonaro. In Venezuela and India, by contrast, states that were both institutionally weaker and politically more vulnerable could not forestall a descent into competitive authoritarianism.

The Limits of the Second Security

Since the 1990s, most instances of transition from democracy to autocracy have occurred not through coups but through incremental erosion of democratic institutions and consolidation of power by elected autocrats (Bermeo 2016; Frantz and Kendall-Taylor 2017). The leaders of such regimes come to power in reasonably free and fair elections, but then set about undermining democratic institutions and expanding their control over the judiciary, the press, the administration of elections, and other institutions of pluralist democracy. Where they succeed in these stratagems, they typically consolidate what scholars refer to as competitive authoritarianism (Levitsky and Way 2010) or electoral autocracy (Shevtsova 2001).

Given these dynamics, one might be tempted to conclude that federalism must, ceteris paribus, make backsliding into autocracy less likely. The reasoning would be simply that because federalism divides authority between two levels of government and then disperses part of that authority geographically across multiple state governments, it must make it more difficult for an aspiring national autocrat to consolidate control. But just as political scientists have shown us that presidential systems with separation of powers do not consistently make the death of democracy less likely (Cheibub 2007; Linz 1990), so too have they shown us that federalism does not consistently function as a safeguard. Indeed, as noted above, a rich literature demonstrates that federalism can encourage the survival of authoritarian enclaves at the state level (Gibson 2012), and that these enclaves may work to spread authoritarianism to the federal level (Bateman, Katznelson, and Lapinski 2018; Grumbach 2022).

But does the fact that federalism does not consistently act as a safeguard of democracy mean that Madison was entirely wrong to suggest it could provide a “second security” for liberty? We argue that federalism might serve as a guardrail against an aspiring autocrat consolidating power at the national level, but—as noted above—only when state governments have institutional resources that allow them to push back against federal encroachment, and when a sufficient number of state governments are controlled by opposition parties or independent factions within the ruling party. We elaborate on each of these factors in turn.

Not all federations are robust (Bednar 2009; 2021). When federal systems lack robustness—for instance, when state governments lack the resources necessary to exercise “meaningful independence” from the national government—federalism will not serve as a guardrail against tyranny (Bednar 2021). Weak states make for weak bulwarks against national autocrats. The more authority state governments have over key government functions and the more resources they control, the more likely they will be able to prevent the consolidation of autocratic power. While no single resource is likely to be determinative, the greater the resources and authority of state governments in areas such as fiscal policy, police powers and the administration of justice, and election administration, the more likely states will be able to constrain backsliding by a national autocrat.

There is of course a great irony here. On the one hand, the more autonomy and authority state governments have, the more likely that some states may shelter “authoritarian enclaves” at the state level that can undermine national democracy (Gibson 2012; Grumbach 2022; Mickey 2015). But on the other hand, as Bednar (2021) notes, the same state strength that can diminish a federal system’s democracy at the state level may protect it from autocratic assaults at the national level.

The distribution of partisan control over state governments is the second important determinant of the effectiveness of federal guardrails. When Madison speculated in The Federalist No. 51 that “[t]he different governments will control each other …,” he was writing in an era before the emergence of modern political parties. But partisan blood often proves thicker than institutional water (Bulman-Pozen 2013). Even if state governments have control over the resources necessary to counter attacks on democracy from an aspiring autocrat at the national level, this will do little to impede backsliding if that leader’s allies control a vast majority of state governments. In some cases, autocratic initiatives at the national level may be resisted by locally based politicians within the aspiring autocrats’ party. But the likelihood of this type of intraparty opposition within a federal system diminishes as party competition becomes more polarized (Bednar 2021) and as politicians and voters become more motivated by “negative partisanship.”

Ultimately, the effectiveness of federalism as a guardrail will depend on whether opposition parties or factions of the ruling party that remain independent of the aspiring autocrat control enough states to resist autocratization. The exact number of state governments that must remain independent of the aspiring autocrat will vary across countries depending on the specific rules governing elections and lawmaking. Moreover, opposition parties might well encounter significant coordination problems as they seek to resist the aspiring autocrat. But generally, we can expect that the more state governments remain under the control of opposition parties, the more likely federalism can help to forestall a transition to autocracy.

Quantitative Analysis of the Relationship between Federalism and Democratic Backsliding

In this section, we assess the relationship between federalism and four main outcome measures: (1) whether backsliding occurs, (2) the severity of backsliding, (3) the speed of backsliding, and most importantly, (4) whether backsliding leads to a transition from democracy to autocracy. We consider the entire population of democracies from 1974 (when the “third wave” of democratization began) through 2021.

Our measurement of democracy and democratic backsliding employs the strategy developed by Haggard and Kaufman (2021a; 2021b), described in detail in appendix 1, using V-Dem data (Coppedge et al. 2022; Pemstein et al. 2022). The universe of democratic regimes is comprised of all countries that experienced at least eight consecutive years with a score of at least 0.5 on the V-Dem electoral democracy index (EDI) (a measure that taps into “the basics” of democracy: free, fair, and competitive elections). ³ Our sample consists of 105 democracies in this period (1974–2021), 18 of which are federal. ⁴

Next, we identify which of these 105 democracies experienced backsliding, using a two-step procedure. In step one, we identify countries that experienced any episode in which their score on the V-Dem liberal democracy index (LDI) fell significantly below the peak LDI score achieved prior to the backsliding episode. ⁵ In step two, we triangulate our measure with other measures of democracy: we examine whether countries identified as backsliders in step one are also identified as eroding toward or reverting into authoritarian rule by at least two other democracy indices, and we only use those that are validated in our final list of backsliding cases (see appendix 1 for the list of countries and appendix 7 for the visualization of backsliding episodes).

Based on our measurement strategy, we identify 21 democratic countries that experienced episodes of backsliding. We find no statistically significant relationship between federalism and the occurrence of democratic backsliding (our first outcome measure). ⁶ As table 1 shows, 22% of the 18 federal democracies have experienced backsliding, while 20% of the 87 unitary democracies have. Though one must be cautious about making inferences from such a small and uneven universe of cases, these results provide no reason to conclude that federalism systematically impedes backsliding.

Table 1 Federalism and Democratic Backsliding

Even if federalism does not make backsliding less likely, one might expect it to restrain the severity of backsliding (our second outcome measure) or the speed of backsliding (our third outcome measure). However, we find no statistically significant relationship.⁷ As we discuss in detail in appendix 3, the severity of backsliding (as measured by taking the difference between a country’s LDI score prior to the first year of the backsliding episode and the lowest score during the episode) is higher in federal systems than in unitary systems. Likewise, the speed of backsliding (as measured by dividing the score for severity by the total years of the backsliding episode) is also higher in federal systems.

Finally, we assess whether, among those democracies that do experience backsliding, federal systems are less likely than unitary systems to descend from democracy all the way into autocracy (our fourth outcome measure). We find that the percentage of federal backsliders that falls below our democratic threshold of 0.5 on the EDI is slightly lower than reversions in unitary systems (50% versus 53%), but this difference is not statistically significant. ⁸ In appendix 4, we also explore whether backsliding is influenced by the degree of decentralization, rather than federalism per se, employing different dimensions of decentralization using the regional authority index (Hooghe et al. 2016) as predictors in a series of logistic regression models. As with the more restricted definitions of federalism, however, we again find no statistically significant relationship to democratic backsliding.

Structured, Focused Comparisons

While the descriptive statistics reviewed above reveal no relationship between federalism and backsliding, that does not mean federalism can never play a role in preventing the demise of democracy. This depends on the institutional capacities of the state governments and on the extent to which they are controlled by parties or leaders opposing the aspiring autocrat. To examine the effects of these factors, we present comparative case studies of the only four federal systems in our sample that experienced significant threats to democracy from above: the United States under Donald Trump (2017–21), Brazil under Jair Bolsonaro (2019–23), Venezuela under Hugo Chávez and Nicolás Maduro (1999 to present), and India under Narendra Modi (2014 to present).

Notwithstanding important differences in social equality and civil liberties across these four countries, each of them had relatively robust national institutions in the period preceding their backsliding episode. There was considerable overlap between the United States, Brazil, and India in terms of V-Dem scores for the autonomy of electoral institutions; in fact, the latter two countries scored higher than the United States throughout most of the 2000s. Not surprisingly, the United States had the highest score for judicial independence, but scores for Brazil and India were relatively close. Venezuela had somewhat lower scores on both measures, but they were still roughly comparable to Brazil and India in the years preceding the election of Hugo Chávez in 1998.

At the same time, as we will see in the structured, focused comparisons below, there were substantial differences in the strength and political control of state-level institutions—the two conditions we argue are necessary for federalism to serve as a guardrail for democracy. In Venezuela and India, state-level financial and legal institutions were weak relative to the national government, and both were vulnerable to control by autocratic leaders at the national level. In both cases, the backsliding process resulted in the countries descending into autocracy—though to date the descent has gone much further in Venezuela than in India.

By contrast, the two key conditions we identify did hold to a much greater extent in the recent episodes of backsliding in the United States and Brazil. Notwithstanding the weakening of democratic institutions, backsliding did not lead to a reversion to autocracy in either case (see figure 1 below). This by no means assures us that these systems are immune from autocratic takeovers in the future; indeed, the institutional bulwarks of democracy remain under threat in both countries. But as we show below, robust federalism did provide important brakes on these threats during the Trump and Bolsonaro presidencies.

Figure 1 Movement in V-Dem LDI across Cases

Notes: The dashed vertical line indicates the starting year of the backsliding episode based on the LDI. The solid vertical line indicates a decline below 0.5 on the EDI score during the backsliding episode, our measure of transition into autocracy.

Source: Coppedge et al. 2022.

The United States

As Lisa Miller (2023) explains, “constitutional folk wisdom” in the US celebrates federalism as one of several virtuous checks and balances that help to protect vulnerable political minorities and prevent tyranny. But as she and many other political scientists have shown, this narrative ignores the substantial antidemocratic features of American federalism, in particular its contribution to racial authoritarianism (Gibson 2012; Grumbach 2022; Mickey 2015; Miller 2023; Riker 1964). Federalism has also provided these autocratic state-level regimes with pathways to infiltrate the federal government (Bateman, Katznelson, and Lapinski 2018). But having recognized the ways federalism has undermined American democracy in some periods, can we nevertheless identify circumstances in which—more in line with Madisonian thinking—it can help to prevent American democracy from sliding entirely into autocracy? We think that the strength and political autonomy of American federal institutions were manifest in the context of the 2020 election, and we argue that these features of American federalism played an important role in preventing President Trump and his coconspirators from subverting American democracy (see also Landau, Wiseman, and Wiseman 2021, who make a similar argument).

As many scholars have noted, Trump and his associates waged a multipronged campaign to undermine America’s democratic institutions during his years in office (Kaufman and Haggard 2019; Lieberman et al. 2019). The impact of this campaign is reflected in the substantial decline in US rankings on many democracy indicators during the Trump years (Boese et al. 2022, 37; see also figure 1 above). Trump’s efforts to dismantle American democracy intensified in the run-up to the 2020 election and reached their apogee in its aftermath.

The Trump administration and its Republican backers at the state level engaged in widespread voter suppression, in particular aimed at voters in communities of color (Ray 2020). When these efforts failed to tip the election in Trump’s favor, he refused to accept the result and orchestrated a multifaceted criminal conspiracy to overturn it (Broadwater, Feuer, and Haberman 2022). Important details of this conspiracy have been laid bare by the December 2022 report of the US House of Representatives Select Committee to Investigate the January 6th Attack on the United States Capitol (2022) and the August 2023 criminal indictment against Trump (United States v. Donald Trump 2023).

We argue that federalism was in fact crucial to the capacity of American democracy to survive these threats. As Landau, Wiseman, and Wiseman (2021, 3) put it, “The 2020 election process demonstrated that state and local governments … served as important (albeit highly imperfect) bulwarks against the threat of authoritarianism. If the federal government had more direct control over elections, the situation—perilous as it was—would likely have been far worse.” ⁹

Federalism played this role during the Trump years because (1) state governments had robust control over key government functions, in particular state judiciaries and the administration of federal elections, and (2) a critical mass of those state governments were controlled either by Democrats who actively opposed Trump’s authoritarian agenda or by Republicans who maintained sufficient distance from Trump to resist supporting his efforts to overturn the election results. As a result, despite their best efforts, Trump and his associates were unable to rig the election decisively in his favor, or to overturn its result.

The highly decentralized administration of elections in the US prevented Trump from throwing the election in his favor. Though some state governments controlled by his partisan allies engaged in voter suppression efforts designed to aid his reelection efforts, too many states remained outside his grasp for him to ensure victory. As Landau, Wiseman, and Wiseman (2020, 1209) note in their work on “partisan federalism,” “[E]ven if states controlled by the in-power party are useless as checks against federal overreach, those controlled by the out-of-power party will have great incentives to oversee the federal government and to check its abuses.”

As figure 2 shows, Republicans had outright control of state government (including both the executive and legislature) in only 22 states ¹⁰ (which collectively amounted to 219 Electoral College votes—less than the 270 needed to win; see table A9 in appendix 5 for the underlying data). Power was divided between Democrats and Republicans in another 13 states (amounting to 124 Electoral College votes), and Democrats controlled 15 others (amounting to 191 Electoral College votes). ¹¹ In several of the key battleground states that proved decisive for the election outcome—including New Hampshire, Michigan, Wisconsin, Nevada, Minnesota, and North Carolina—Democrats held enough power in state government to block efforts by Trump and his associates to decisively rig the election in his favor or overturn the results once he lost. Even in some states controlled overwhelmingly by Republicans, some Republican election officials remained sufficiently independent of Trump and true to the principles of democracy that they refused to do his bidding when he sought to overturn the election. Most famously, Georgia Secretary of State Brad Raffensperger resisted pressure from Trump to “find” him the 11,780 votes he needed to overturn his election loss (Amy 2021; Amy, Superville, and Brumback 2021).

Figure 2 Partisan Control of States in the United States, November 2020

Just as state control over electoral administration made it more difficult for Trump to throw the election in his favor, the decentralization of judicial authority in the US frustrated his efforts to overturn Biden’s electoral victories. In the aftermath of Biden’s victory in the 2020 presidential election, Trump’s allies filed a flurry of lawsuits challenging the results, most of them in the six battleground states of Arizona, Georgia, Michigan, Nevada, Pennsylvania, and Wisconsin. By the end of 2020, Trump and his allies had filed a total of 62 election lawsuits and had lost 61 of them. ¹² As the US Supreme Court refused to take up Trump lawsuits seeking to overturn the election, adjudication of these cases was left to lower federal and state courts (Cummings, Garrison, and Sergent 2021). Though Trump and his allies were unsuccessful in both federal and state courts, the vast majority of his litigation—and thus of his losses—occurred before state courts (Cummings, Garrison, and Sergent 2021; Feuer 2020; Long and White 2020; Wheeler 2021). For instance, “State Supreme Courts in Arizona, Nevada, and Arizona each rejected or declined to hear Trump’s appeal to overturn results in those states, while the Pennsylvania and Michigan supreme courts denied multiple lawsuits” (Cummings, Garrison, and Sergent 2021). Wheeler’s (2021) close scrutiny of votes by individual judges on multi-judge panels shows that Republican-affiliated judges were far more likely to support Trump’s claims than Democrat-affiliated judges. However, his analysis also shows that only half the votes cast in these election cases came from judges with known Republican affiliations, and even these Republican-affiliated judges rejected Trump’s claims nearly two-thirds of the time. If we imagine a counterfactual in which the US judiciary were far more centralized, it is not difficult to imagine scenarios in which Trump and his allies might have gained control over the courts prior to the 2020 election, and then convinced them to overturn the election results. Indeed, focusing on judicial capture has been a central stratagem of many aspiring autocrats (Levitsky and Ziblatt 2018). However, the power of states in election administration, and the power of state courts (which Trump could not control) in adjudicating election challenges, frustrated his efforts to overturn the 2020 election.

Developments in 2020 underlined an irony of American federalism highlighted by Jenna Bednar (2021, 5): the same decentralization of election administration that has so often allowed local authoritarian enclaves to emerge seems to have served to protect American democracy when an aspiring autocrat sought to consolidate power at the national level. Unlike the Venezuelan or Indian cases discussed below, in 2020, the US satisfied the two necessary conditions we identify above for federalism to have a constraining effect on democratic backsliding: (1) state governments had robust control over key government functions and (2) a critical mass of state governments was controlled by parties that opposed the aspiring national autocrat. To be clear, we do not take this finding as grounds for complacency. Though key conditions were met and federalism frustrated Trump’s efforts to subvert American democracy in 2020, there is no guarantee such conditions will be met in the future. Indeed, there is evidence that Trump loyalists are attempting to undermine these very safeguards in hope of restoring him to power in 2024 (Berzon and Corasaniti 2024; Daley 2022; Stone 2023).

Brazil

Brazilian democracy also survived autocratic threats from its previous president, Jair Bolsonaro—a self-styled “Trump of the Tropics”—who occupied the presidency from January 2019 to 2023. As in the United States under Trump, the quality of democracy deteriorated significantly, both in the years preceding the rise of Bolsonaro and during his term in office. Between 2011 and 2021, according to V-Dem’s “Regimes in the World” classification, Brazil declined from a “liberal” to an “electoral” democracy; that is, a system with free electoral competition but one that is deficient in horizontal checks and the rule of law. But like the United States, most observers agree that the regime in Brazil in remained democratic (Boese et al. 2022, 23). In the highly polarized presidential election of October 2022, Bolsonaro and his allies were forced to accept defeat by his center-left opponent, Luiz Inácio Lula da Silva. In this section, we look first at the threats posed by Bolsonaro to Brazil’s democracy and then assess the role played by the federal system in thwarting them.

Backsliding in Brazil and the Rise of Bolsonaro

The road to Bolsonaro’s rise to power was paved by a confluence of crises that severely impacted support for Brazil’s political institutions. Between 2014 and 2016, a major economic recession hit Brazil and a massive corruption scandal engulfed much of the political class, including—although not limited to—leaders of the governing Workers’ Party (the PT). Luiz Inácio Lula da Silva, the popular former president, was caught up in the scandal, and was imprisoned for over a year based on what later was shown to be a highly politicized prosecution. Although the scandal did not directly implicate Dilma Rousseff, his handpicked successor, she was impeached by opposition legislators for abuses of fiscal authority and removed from office. Rousseff’s successor as president (2016–18), Vice President Michael Temer, leader of the Brazilian Democratic Movement Party (PMDB), was also briefly arrested on corruption charges.

Jair Bolsonaro thus emerged from the fringes of politics in a dysfunctional system in which the leaders of the major political parties were already discredited. Though he had been a member of the Chamber of Deputies since 1990, he had remained at the margins of the system—noted mainly for extremist views regarding homosexuality and his nostalgic praise of Brazil’s former dictatorship—until the meteoric rise he experienced in his 2018 presidential campaign. In October 2018, he decisively defeated the candidate of the weakened PT and assumed the presidency in January 2019. Bolsonaro’s core support during the presidential campaign came from evangelicals and lower ranks of the military and police, but he also capitalized on widespread antipathy toward the PT among middle-class voters in the southern states and among business elites, and growing public disillusion with political institutions (Hunter and Power 2019, 74). As president, he appointed military and far-right political allies to cabinet-level positions, engaged in assaults on the civil rights of Afro-Brazilians, the LGBTQ+ community, and Indigenous populations, and undermined restrictions on logging and mining in the fragile Amazon region. This behavior raised serious concerns about the future of Brazilian democracy.

Along with these patterns of norm breaking, several controversies raised more immediate concerns in the final two years of Bolsonaro’s term. First, tensions between the president and the Supreme Federal Court (STF) escalated. Bolsonaro increased his threats to ignore rulings, attacked Justice Alexandre de Moraes, who was scheduled to supervise the 2022 election, and encouraged mass protest rallies against the court. Serious questions also arose about the constitutional loyalty of Bolsonaro supporters in the state-level military police, with many participating in pro-Bolsonaro demonstrations in the months preceding the 2022 vote.

Finally, following the playbook adopted by Trump in the run-up to the 2020 presidential elections in the United States, Bolsonaro preemptively warned of “corruption” in the electoral system, sending strong signals that he would not peacefully accept an electoral loss in 2022 (Horton and Gragnani 2022). When Bolsonaro lost narrowly in the October 30 runoff election, activists reacted with charges of fraud, blockades of major highways, and protests in front of military bases. Lula peacefully assumed office on January 1, but on January 8, the protest climaxed when a mob stormed government offices in Brasília. Although Lula was not in the capitol and Congress was in recess, the mob engaged in widespread destruction before being expelled by security forces at the end of the day.

Checks on Bolsonaro

Many factors account for the survival of Brazilian democracy during the Bolsonaro era, but robust federalism—a feature of the system dating back to the late nineteenth century—was clearly one of them. Governors played a pivotal role in Brazil’s transition to democracy in 1985 and were arguably dominant in the decade immediately following the transition (Abrucio 1998; Samuels 2000; Samuels and Abrucio 2000, 46). National political elites and institutions acquired greater agenda-setting power after the late 1990s (Cheibub, Figueiredo, and Limongi 2009; Hagopian, Gervasoni, and Moraes 2009), but Brazil remained a highly decentralized system. Compared to Venezuela and other federal systems in Latin America, subnational governments in Brazil maintained the highest levels of both expenditures and tax revenue in the region, and governors exercised extensive authority over education systems, the provision of healthcare and other public goods, and subnational appointments (Falleti 2005, 334–35; Fernandes and Santana 2018, 4). All of this provided governors with substantial patronage resources that could be deployed in both local and national electoral contests and allowed them to retain considerable leverage within Brazil’s fragmented multiparty system.

As in the United States, moreover, judicial authority is shared between state and federal courts, and governors also have direct authority over the military police, who are responsible for maintaining public order and can be deployed to handle major security crises. As we discuss below, concerns about the constitutional loyalty of these forces mounted during the Bolsonaro administration, but they did not pose serious threats to the conduct or outcome of the 2022 presidential elections. As in the United States, moreover, the Senate provided additional leverage for the projection of local power centers into the national political arena; most recently, it placed an important constraint on Bolsonaro’s demand for the impeachment of several STF justices.

Unlike in the United States, Brazilian elections are overseen and adjudicated by a central electoral authority (the Superior Electoral Court, or TSE) and the STF, which have the authority not only to regulate and administer elections, but to fine or deplatform inflammatory disinformation posted on social media. Potentially, we have suggested, this concentration of electoral authority might have exposed it to political capture by an aspiring national autocrat, as had occurred under Chávez in Venezuela. In Brazil, however, this would have been very difficult to do, at least in the short term. Only two of the seven members of the TSE are appointed by the president, who must choose from nominees selected by the STF; the others are elected by secret votes from the STF and the Superior Court of Justice. In turn, the decentralization of the Brazilian federal system constitutes a major impediment to changing these institutions. A successful challenge by an autocrat would require forging disciplined legislative majorities or supermajorities—a major hurdle in Brazil’s fragmented multiparty system.

The constraints of the federal system were evident from the outset of the Bolsonaro presidency. Among the governors elected in 2018, candidates who were at that time aligned with Bolsonaro won 13 of 27 governorships, including governorships in three of the wealthiest states—Rio de Janeiro, São Paulo, and Minas Gerais. But eight governors were allied with the opposition, and six were unaffiliated. Figure 3 displays the partisan control of states in the aftermath of 2018 gubernatorial elections (see table A10 in appendix 5 for the underlying data).

Figure 3 Partisan Control of States in Brazil, October 2018

Just as important, the pro-Bolsonaro alliance of governors proved unstable and precarious. Especially bitter conflicts erupted with João Doria from São Paulo. Doria had attached himself to the Bolsonaro bandwagon in 2018, but by 2021 had become widely discussed as a potential rival for the presidency. In Rio, Bolsonaro’s ally Wilson Witzel broke with Bolsonaro over the latter’s refusal to address the COVID-19 pandemic, and in any case was suspended from office in August 2020 over corruption charges. More generally, the majority of governors resisted Bolsonaro’s efforts to minimize the seriousness of the COVID-19 crisis and imposed restrictions despite Bolsonaro’s opposition.

Governors also pushed back strongly against Bolsonaro’s attempts to undermine their authority over the state police. Members of the state police did join protests and rallies in the run-up to the 2022 election, defying legal prohibitions against such activities. But 25 of Brazil’s 27 governors issued a statement in 2021 reaffirming their commitment to “ensure that the mission of the state police takes place within constitutional and legal limits” (Paraguassu 2021). As noted above, there was only limited police participation in postelection protests.

Constraints on Bolsonaro related to federalism were evident as well in the national legislature, where increasing party fragmentation both reflected and expanded the space for particularistic local interests. Between 2010 and 2018, the vote share of the three “national” parties (the PT, PMDB, and the Brazilian Social Democracy Party [PSDB]) declined from about 30% to 16.5% of the vote, while the number of legislative parties—most with strong state and municipal roots—grew from 26 to 31. The government’s congressional coalition was initially composed of 11 different parties. Bolsonaro himself quickly broke with the small party that nominated him (the Social Liberal Party [PSL]) and governed as an independent for much of the rest of his term.

As noted in the discussion of the TSE above, the lack of a cohesive ruling party—itself a reflection of the impact of state-level political machines—severely limited opportunities for Bolsonaro to expand control over the states, courts, or other independent centers of power. For the first two years of his term, Bolsonaro engaged in a contentious relationship with Rodrigo Maia—the speaker of the Chamber of Deputies with extensive authority over the legislative agenda. In February 2021, Bolsonaro engineered the election of a political ally, Arthur Lira, as the new speaker, and bolstered his legislative support by appointing members of his coalition into cabinet positions. This helped to insulate him from impeachment, but the support depended on the backing of the “Centrão,” a loose alliance of center-right politicians primarily interested in access to local patronage and pork (Álvarez and Savaresef 2021). That support would melt away in the aftermath of the 2022 election.

On October 30, 2022, Bolsonaro lost the runoff election by only 1.8% of the vote and on January 1, Lula assumed the presidency. Given Brazil’s highly polarized political environment, it is not surprising that the transition was fraught. As noted above, Bolsonaro supporters reacted with highway blockades, demands for military intervention, and the storming of government offices in Brasília.

Yet Brazilian institutions held. Immediately after the election, the results were acknowledged by powerful national and subnational officials, including most of Bolsonaro’s congressional and gubernatorial allies. In the aftermath of the attack, as previously noted, questions emerged about the loyalty of the security forces protecting the capitol and about the complicity of Ibaneis Rocha, the governor of the Federal District, who was suspended for 90 days by the head of the STF. Nevertheless, all of the governors (including Rocha), as well as national leaders across the political spectrum, again affirmed the results of the election and publicly condemned the mob violence. Unlike the US capital riots on January 6, 2021, the attack had posed only a limited threat to the transfer of power.

It is important to note that the democratic transfer of power in Brazil cannot be attributed entirely to the federal features of the constitution. On the contrary, national and international actors played a major role, including the military establishment, which proved unwilling to challenge the results. Indeed, the most direct defense of the electoral process came from the STF and the TSE themselves, which quickly ratified the results of the election and foreclosed the long delays and litigation that had accompanied the electoral process in the United States. As noted above, however, the independence of these agencies rested on the centrifugal pull of parties that retained strong local roots. More generally, to prevail against the prodemocratic forces operating at the national level, Bolsonaro and his allies would have had to find ways to gain control of the dispersed centers of institutional power that have long characterized the Brazilian political system.

As in the United States, the strains on Brazilian democracy can be expected to persist. Brazil remains a highly polarized political system, but thus far the democratic system has survived.

Venezuela

The weakness of federal checks on backsliding in Venezuela stands in marked contrast to the role of state actors and institutions in the United States and Brazil. From 1989 to 1993, prior to the election of Hugo Chávez, successive Venezuelan governments instituted a series of decentralizing reforms; the most important of these was a provision for the direct election of governors and mayors, who had previously been appointed by central authorities. But although this change had an important short-term impact on party politics, the decentralization of other governance functions was much more limited than in the United States and Brazil. State-level authorities in Venezuela lacked control of the electoral, judicial, and administrative resources that provided opportunities for resistance to backsliding central governments headed by Trump and Bolsonaro. For this reason, the reforms of the late 1980s and early 1990s did relatively little to impede the centralization of power engineered under Chávez after 1998.

Centralization and Decentralization, 1958–93

Although Venezuela has long been formally constituted as a federal system, in practice, authority has been centralized almost completely at the national level throughout most of the country’s democratic experience. Governors and mayors were appointed by the president. The national government retained control of the judicial system and the electoral authority as well. And the political power of the dominant party elites was consolidated by a closed-list proportional representation system that gave their central executive committees control of nominations for Congress, state assemblies, municipal councils, and judges.

Pressures to decentralize the system began to build during the 1980s, reflecting widespread public dissatisfaction with the two-party duopoly of the dominant Democratic Action (AD) and Social Christian (COPEI) parties. The old guard of both parties, predictably, resisted decentralization reforms, but their hold was weakened in 1988 by the election of Carlos Andrés Pérez, who had won the AD nomination with grassroots support, and then by the eruption of widespread social protest and police violence that broke out in Caracas in 1989.

As we will see in more detail below, the limited reforms that followed over the next several years left major governance functions in the hands of the federal government. But as a step toward restoring the legitimacy of the system, congressional majorities agreed to establish a mixed electoral system and the direct election of governors and mayors, and this, as Michael Penfold-Becerra (2004, 197) writes, “changed the field of political action.” The AD and COPEI fractured in the face of strong local challenges to the national leadership, and new political opportunities at the state and municipal levels opened the way to the proliferation of regionally based parties, which by 1998 controlled 10 of the 23 governorships in the country. The fragmentation of the party system also opened the way to the rise of outsider challenges for the presidency. Over the course of the 1990s, support for the dominant parties evaporated. In 1994, Rafael Caldera, formerly a stalwart of the COPEI, won the presidential election running as an independent. In 1998, the victor was Chávez, the military leader of a coup attempt in 1992, and his main challenger was Henrique Salas Römer, who had previously been elected governor of the state of Carabobo as an independent.

The Chávez Regime (1998–2013): The Centralization of Power

Chávez’s victory in the 1998 presidential election marked a critical turning point for Venezuelan democracy. He had campaigned with a promise to overhaul the existing system, and he capitalized on his political honeymoon to elect a Constituent National Assembly that was dominated overwhelmingly by his supporters. Predictably, the assembly produced a constitution that vastly increased the powers of the president and correspondingly weakened horizontal checks at the national level. It extended the presidential term from five to six years, with the possibility of immediate reelection, gave him the authority to appoint directors of the Central Bank, and established a unicameral legislature dominated by the ruling coalition. With support from the dominant legislative coalition, in turn, Chávez could wield extensive decree powers with respect to hydrocarbons and agriculture. Unlike Bolsonaro in Brazil, Chávez’s reliable legislative majority also made it “easier for the president to pack the court and tighten control over the attorney general, the comptroller general, and the [National Election Council], … the electoral monitoring body” (Corrales and Penfold 2011, 122).

Notwithstanding the collapse of the separation of powers among national institutions, the decentralization of the electoral system did initially provide some basis for independent political forces. Regionally based parties within Chávez’s own coalition supported constitutional reforms that increased governors’ terms from three to four years and provided for nonconcurrent elections. And although Chávez’s ruling party won a large majority of the governorships, state-level elections provided an institutional foothold for opposition challengers. Carabobo—the principal site of the petrochemical industry—provided the basis of Salas Römer’s challenge to Chávez in the 1998 election. Francisco Arias, the candidate in 2000, and Manuel Rosales, the candidate in 2006, were both governors of Zulia, the wealthiest state in the country. Henrique Capriles, who ran against Chávez in 2012 and against Maduro in 2013, was governor of another relatively wealthy state, Miranda, between 2008 and 2017. As figure 4 shows, although Chavista forces dominated after 1998, opposition governors continued to control at least some of Venezuela’s states throughout the Chávez period (see table A11 in appendix 5 for the underlying data).

Figure 4 Partisan Control of States in Venezuela, 2000–16

Why were these candidates unable to leverage their regional bases in relatively wealthy and powerful states into a challenge that might have halted the descent into authoritarianism? First, as has been widely analyzed elsewhere, political challengers faced powerful political and economic headwinds at the national level, particularly after the onset of the oil boom in 2003. The opposition itself, moreover, was divided and discredited by association with the old “Punto Fijo” regime. Although governors did form a multiparty association to defend state interests (Penfold-Becerra 2004), differences in oil wealth, industrialization, and geographic location proved a serious impediment to collective action (Eaton 2014, 1147).

But the weakness of federal checks was also attributable, at least in part, to the limited nature of the earlier decentralization reforms, which left state governments highly vulnerable to the expansion of national power under Chávez. While the election of governors provided some opportunity for competing politicians to build a political base, they lacked control over core features of governance characteristic of the more robust federal systems in the United States and Brazil.

First, although fiscal transfers increased somewhat over the course of the 1990s, state governments remained almost entirely dependent on the national government for revenue. Chávez, in turn, deployed oil revenues and restructured the value-added tax to redirect the flow of resources toward social welfare programs and communal councils controlled by the national government (Eaton 2014, 1146). Second, by 1998, only a few states (Aragua, Lara, and Nueva Esparta) had acquired administrative responsibilities for education, and in 2001, Chávez’s congressional majority granted him sweeping new decree powers over the educational system. The decentralization of healthcare, although more extensive than education, also remained incomplete (Penfold-Becerra 2004).

Finally, states acquired virtually no authority in any of the functions of governance that might have enabled them to resist Chávez’s consolidation of autocratic power. The judicial system and electoral agencies remained entirely controlled at the national level, and vulnerable to political pressure and reorganization under Chávez. As early as 1999, the Supreme Court yielded to demands that legislative authority should be transferred from the sitting Congress, where opposition parties still had veto power, to the Constituent National Assembly, where Chavistas had overwhelming dominance. In 2000, with a new unicameral legislature in place, a judicial reform commission dominated by Chávez loyalists authorized a replacement of controversial judges with Chávez allies. And in 2005, following a victory in a recall referendum, the National Assembly approved a new “Organic Law of the Supreme Court,” which allowed the Chavista legislative majority to appoint and remove judges and to expand the number of seats from 20 to 32 judges, assuring a full subordination of the judicial system.

The National Electoral Council (CNE) suffered a similar fate. Throughout the early 2000s, it had been compromised by the appointment of Chavista loyalists, and in the run-up to the presidential election of 2006, the ruling party in the legislature appointed an entirely new board of directors, cementing the government’s control. Although the election was generally considered free and fair, Freedom House noted that the CNE was ineffectual at limiting Chávez’s massive advantages in the use of public television, infrastructure investment, or the promotion of social projects.

In subsequent years, as opposition forces began to mobilize more intensively, Chávez wielded the power of the national government even more directly against his political opponents. In the run-up to the 2008 state and local elections, the CNE disqualified over 300 opposition candidates on charges of corruption. State prosecutors forced Manuel Rosales, Chávez’s opponent in the 2006 presidential elections, into exile on corruption charges after he was elected mayor of Maracaibo in 2009. Also in 2009, the government pushed through a referendum lifting term limits after it had gone down in a surprise defeat in 2007. And in 2010, as a unified opposition coalition organized to challenge the control of the ruling United Socialist Party of Venezuela (PSUV) in Congress, the PSUV changed electoral laws to ensure that it could continue in the majority. Its candidates captured 48% of the vote, compared to 47% for the opposition, but it secured 98 of the 165 seats.

Central to these efforts was national control over the prosecutorial apparatus and law enforcement. Though states and municipalities did perform some local police functions, national agencies controlled much of the policing functions even before Chávez came to power. The National Guard, with approximately 33,000 officers, was responsible for ports, freight and contraband, and prison security, as well as for “public order” (Birkbeck 2009, 288). A separate force (the Scientific, Penal and Forensic Investigation Corps), responsible for criminal investigation, worked closely with the public prosecutor’s office. An intelligence service attached to the Ministry of Interior was responsible for national security and counterespionage, and a fourth branch, with about 5,800 officers, had primary responsibility for traffic policing.

During the early 2000s—a period when Chávez faced highly mobilized opposition from business elites and civil society groups—both the police and the military divided along partisan lines in confronting protests and demonstrations. The government “us[ed] the National Guard and pro-government state and municipal police forces to ‘protect’ marches in favor of Chávez and ‘control’ opposition marches, while opposition state governors and municipal mayors did the reverse” (Birkbeck 2009, 297). In 2002, an opposition faction within the armed forces held Chávez prisoner for several days before support for the coup collapsed and he was restored to power.

Chávez’s ability to survive these challenges rested on a variety of factors, including social welfare programs financed by the onset of the oil boom in 2003, and the questionable legitimacy of opposition forces associated with the old regime. But central control of the state’s coercive apparatus clearly helped to tilt the odds in his favor. Police forces controlled by opposition governors were no match for the National Guard, and reorganizations undertaken in the mid-2000s reinforced the authority of the Ministry of Interior. In the aftermath of the aborted 2002 coup, Chávez also thoroughly purged the military establishment and organized a new, national militia force and neighborhood “communes” as counterweights.

By 2005–6, Chávez had succeeded in consolidating power. He had weathered the strikes and demonstrations of the early 2000s. He had faced down a coup attempt in 2002, a crippling strike in the petroleum sector in 2003, and an attempt to oust him through a recall referendum in 2004. The opposition, conversely, was demoralized and divided. Among other things, it won only two state governorships in the 2004 gubernatorial elections (see figure 4), and its decision to boycott congressional elections in 2005 left Chavista forces in full control of the national legislature, opening the way to an extension of controls over the economy, civil society, and the press. Although national and local challenges continued—and even escalated after Chávez’s death—neither the federal system nor other impediments had prevented a long slide into authoritarianism.

India

Since his election as prime minister in May 2014, Narendra Modi and his Bharatiya Janata Party (BJP) government have waged a systemic, multifaceted campaign to undermine India’s democracy (Khaitan 2020; Vaishnav 2021; Varshney 2022). Country experts continue to debate how India’s governing regime should be categorized (Dobson and Masoud 2023), and some (e.g., Varshney 2022) do not think India should yet be categorized as an electoral autocracy. However, by 2019, the year Modi was elected to a second term, according to our measurement strategy for identifying electoral autocracies (following Haggard and Kaufman 2021a; 2021b), India had descended from the category of democracy to the category of electoral autocracy or hybrid regime. Leading ratings bodies reach similar judgments. V-Dem’s 2022 report downgraded India from an electoral democracy to an electoral autocracy (Boese et al. 2022), and Freedom House has lowered India’s freedom score from “free” to “partly free” (Freedom House 2022). As Maya Tudor (2023) put it recently, “India exemplifies the global democratic recession.”

Backsliding under Modi (2014–Present)

The general elections of 2014 put an end to the rule of coalition governments in India that had prevailed for almost two decades, as Modi’s BJP won a parliamentary majority. This reinstated one-party dominance at the center, which India had not seen since the 1980s when the Indian National Congress party had last dominated single-party majority governments. Once in power, Modi’s BJP government undertook a wide-ranging series of actions designed to eliminate checks on executive authority, pursue its Hindu nationalist agenda, and ultimately undermine Indian democracy itself. The Modi agenda amounted to what Tarunabh Khaitan (2020, 49) has called “killing a constitution by a thousand cuts,” with an approach that was “subtle, indirect, and incremental, but also systemic” (see also Freedom House 2022; Vaishnav 2021; Varshney 2022).

Reforms and actions the Modi government and its allies undertook included changing the timing of elections and campaign finance rules in ways that favored the ruling BJP; attacking the fundamental rights and equality of Muslim citizens; using legislative chicanery to circumvent the upper house’s (Rajya Sabha’s) veto; attacking critical civil society organizations, including by jailing thousands of activists on sedition charges; threatening the independence of the judiciary; harassing critical journalists; and packing nominally independent bodies (such as the anti-corruption ombudsman [Lokpal] and the Central Bureau of Investigation) (Freedom House 2022; Khaitan 2020; Vaishnav 2021; Varshney 2022).

To be sure, some central institutions have resisted some of the Modi government’s assaults on pillars of India’s democratic order. Despite the great pressures the Modi government has placed on the judiciary, in some instances the Supreme Court stymied his initiatives. One salient example involves the Supreme Court’s reaction to the Modi government’s use of the so-called President’s Rule—a constitutional provision (Article 356) that empowers the union government in exceptional circumstances to suspend state self-government and impose direct union-government rule over the state. This rule is one of the most unitary features of the Indian constitution, as it allows the president to dismiss state governments for partisan purposes. The President’s Rule was rarely used during the era of coalition governments. In 2016, however, the BJP used the President’s Rule to take control of two states—Arunachal Pradesh and Uttarakhand. However, the Supreme Court nullified the application of the President’s Rule on both occasions when Modi’s government sought to invoke it (Sharma and Swenden 2018). More generally, however, existing safeguards of Indian democracy have proved unable to effectively resist the gradual backsliding of the Modi years.

The Failure of Federal Safeguards

Though India clearly has not experienced backsliding as severe as that in Venezuela, it has dropped into the category of electoral autocracies according to V-Dem, and its federal constitutional system has not managed to prevent this. Can our argument explain the failure of federalism to serve as a more effective guardrail of Indian democracy? We think it can.

Unlike the United States or Brazil, but like Venezuela, India does not satisfy the first necessary condition we identify for federalism to act as a safeguard against backsliding—that state governments have robust control over key government functions. Despite some decentralizing reforms in the 1990s, Indian state governments remain weak. Even where states do wield authority, their control is often tenuous, and Prime Minister Modi found avenues to expand his government’s power at the expense of states. The Indian federal system also failed to satisfy our second key condition: by 2018, 21 of India’s 29 state governments were controlled by the BJP or a BJP-related coalition (see figure 5). Consistent with our theory, under these conditions, Indian federalism has not served as an effective bulwark for democracy. Though there certainly has been more pushback by opposition-controlled states against the central government’s autocratic agenda in India than there was in Venezuela, this has not prevented a decline into competitive authoritarianism.

Figure 5 Partisan Control of States in India, 2014–18

State governments were relatively weak in India long before Modi took office, and once in power his government was able to further erode their position, thus preventing them from acting as an effective check on his consolidation of power. India was originally constituted as a federal system as part of a strategy to cope with the ethnic divisions and polycentrism that threatened the unity of the new polity (Tummala 1992). Against this backdrop, the Indian constitution combined federalism with many unitary features and granted strong powers to the center. Under certain circumstances, Parliament had authority to intervene in the affairs of state governments, with the most dramatic example being the so-called President’s Rule mentioned above. There had been some decentralization of power in the 1990s, focused on local government. In part this was due to the rise of coalition governments at the national level—which included several regional parties that demanded such reforms (Sridharan 2003). The liberalization of the Indian economy in the 1990s also encouraged decentralization, as it entailed granting state governments greater powers over economic policies. Nevertheless, the center continued to enjoy a dominant position over states and the foundations of decentralization were fragile (Bagchi 2003).

After Modi came to power in 2014, he not only manipulated the electoral system and waged attacks on independent national institutions, but also attacked federalism itself. The fact that he led a single-party national government and did not depend on coalition support from state-based parties opened up space for him to centralize (Sharma and Swenden 2018, 54). Many scholars note the centralization of powers has considerably transformed Indian federalism since 2014 (Khaitan 2020; Khosla and Vaishnav 2021; Sharma and Swenden 2018).

The Modi government attacked the prerogatives of states in several ways. First, as noted above, in 2016 the BJP attempted to use the President’s Rule to oust opposition governments in the states of Uttarakhand and Arunachal Pradesh (Sharma and Swenden 2018). Though these efforts were thwarted by the Supreme Court, two years later, in a similar scheme, the central-government-appointed governor in the state of Jammu and Kashmir (normally a primarily ceremonial office) dissolved the state house and called for new elections to prevent an alliance of opposition parties from forming a government (Khaitan 2020, 69). Likewise, the Modi government used lieutenant governors—another office appointed by the center—to interfere in the governance of jurisdictions such as Delhi and Puducherry where the opposition controlled the elected governments (70). The central government also used its control over the military and federal police to wield force within the opposition-controlled state of Bengal (71). Finally, in perhaps the most dramatic example of Modi’s attack on state powers, after his reelection in 2019 he dissolved the statehood of Jammu and Kashmir (the only majority-Muslim state in India) and made it a union territory ruled by the center (Varshney 2022, 110).

To be sure, opposition-controlled states have served as important bases of resistance to the Modi government’s autocratic agenda. Opposition parties have continued to score victories in important states such as West Bengal and Punjab. The BJP remains a very weak force in southern states such as Kerala and Tamil Nadu, and in a number of states it is forced to govern in coalition with other parties. Opposition control of several powerful states has certainly slowed the Modi government’s pursuit of its autocratic agenda, for instance when non-BJP-controlled states refused to implement its discriminatory Citizenship Amendment Act and National Register of Citizens (Varshney 2022, 111).

But while opposition-controlled states have resisted Modi’s agenda, they have not been able to block it decisively. The limited powers of state governments discussed above, coupled with the BJP gaining control of government in the majority of states, has meant that state governments could not prevent Modi from decisively eroding Indian democracy. While the BJP controlled just five of India’s states in 2013, by March 2018, it controlled 21 either on its own or in coalition with other parties (Schakel, Sharma, and Swenden 2019). Figure 5 displays the partisan control of states in India in 2014 when Modi came to power and in 2018, shortly before he faced his second general election (see table A12 in appendix 5 for the underlying data). Given the rapid establishment of BJP dominance across so many Indian states, while opposition enclaves could certainly resist the Modi government, they could not harness federalism as a decisive institutional safeguard against backsliding.

Conclusion

Our analysis highlights a powerful irony at the heart of the relationship between federalism and democracy. As Jenna Bednar (2021, 5) notes, “Federalism may be the seed that begins a turn toward democratic backsliding by enabling authoritarianism to take hold in a state. However, in providing an extra layer of diversity, redundancy, and modularization, it may also be a repository of democracy.” Some of the same features of federalism that sometimes diminish and imperil democracy can in other circumstances help to guarantee its survival.

Madisonians and others who place the most faith in federalism as a bulwark of democracy overestimate its impact because they do not recognize that, under many conditions, federalism will not help to safeguard democracy. Systematic, comparative analysis of federal systems around the world suggests that in fact federalism is not consistently a reliable safeguard for democracy. What is more, an extensive literature on subnational authoritarianism cited above emphasizes how federalism can help to perpetuate authoritarian enclaves—which can sometimes poison national democracy as a whole. At the same time, those who dismiss federalism’s role as a bulwark and see it as actually undermining democracy (i.e., Grumbach 2022) do not acknowledge that under specific conditions, federalism may play a vital role in helping to thwart the ambitions of an aspiring autocrat at the national level.

Our analysis disputes more sweeping claims about the role of federalism as a bulwark of democracy and instead advances a more contingent, midrange institutionalist theory (Merton 1968, 39; Steinmo, Thelen, and Longstreth 1992, 11–12) specifying the conditions under which federalism is likely to safeguard democracy. As we argue above, and as we demonstrate in the US and Brazilian case studies, where states control key government functions and opposition parties control a sufficient number of state governments, as they did during the episodes of backsliding in the US and Brazil, federalism may help to prevent an aspiring national autocrat from supplanting democracy. Where these conditions are not met, as was the case in the episodes of backsliding we analyze in Venezuela and India, federalism is unlikely to act as a meaningful impediment to democratic backsliding.