Democratic Stability: A Long View

What are the sources of democratic stability? The evidence from three modern waves suggests that stability rests on economic growth, strong states, and liberal institutions. But can we secure democratic stability beyond liberalism? This question is relevant to those developing countries that have little hope, and perhaps little interest in liberal democracy. But it is also increasingly relevant to those developed nations where the achievements of the twentieth-century liberal order are being eroded. This article takes a fresh look at democratic stability by reviewing the evidence from the last two and a half millennia. Particular attention is devoted to the case of ancient Athens, which highlights the importance of alignment between shared norms and appropriately designed institutions. Athens’ case suggests that goods that we usually associate with modern liberal democracy do not necessarily rely on a given set of values and do not have a unique institutional manifestation.


INTRODUCTION
The institutions and values of liberal democracy, once improbably assumed as the telos of history (Fukuyama 1992), are wearing. 1 The signs abound: the election of Donald Trump in the United States, the Brexit vote in the United Kingdom, the rise of far-right parties across Europe-from neo-Nazis in Germany to populists in Italy-and the authoritarian turn in Turkey and Hungary, to name a few. These events have made questions of democratic stability and breakdown more pressing.
Such questions would have been peculiar a mere 70 years ago, when democracy was still an oddity-as it had been throughout the whole of recorded human history. Through three waves that spanned roughly the length of the twentieth century (Huntington 1991), democracy became the default regime around the world, first in Western Europe, then in Latin America and Asia Pacific, and finally in Eastern Europe and sub-Saharan Africa. As the waves ebbed, we learned more about what makes democracies survive and what dooms them to collapse. As I discuss in Section 2, the evidence suggests that democratic stability rests on economic growth, strong states, and liberal institutions, such as a robust rule of law, free and fair elections, and individual rights. 2 Democratic breakdown occurs instead in the aftermath of military coups (O'Donnell 1973), or when leaders lose legitimacy or cannot solve political problems (Linz & Stepan 1978), and it is more likely under presidentialism (Linz 1990), when inequality is high (Boix 2003, Houle 2009), or when the country has past experiences with authoritarian institutions (Cheibub 2007). Breakdown also occurs through the erosion of checks on elected leaders (Ginsburg & Huq 2018, Levitsky & Ziblatt 2018, Mainwaring & Bizzarro 2019. As challenges to democracy reach western liberal nations, the time has come to take a fresh look at the sources of democratic stability. Because our knowledge depends on evidence drawn uniquely from the modern world, the results may obscure the role of underlying factors associated with modernity-for example, industrialization or military technology. To rule out these factors, we must expand the analysis to the premodern world. This requires loosening the terms of the inquiry, particularly when it comes to the definition of "democracy." 3 But, as I hope to show, if we may lose some analytical rigor, we gain breadth of understanding. For example, the evidence that I discuss in Section 3 suggests that economic growth is important, but modern levels of growth are not necessary for democratic stability. Institutions and rights are also important, but the package 1 In this article, I use "liberal" and "liberalism" in the contemporary sense of institutions and values promoting social justice and universal human rights, not in the classical sense of institutions and values that defend the autonomy and natural rights of the individual against the state. I do not wish to imply, normatively, that liberal institutions and values are not worth promoting or protecting. Instead, this article takes its cue, empirically, from the evidence that these institutions are now eroding, and it turns to history to investigate the possibility of alternative paths to human and societal flourishing. 2 I am concerned here with democratic stability rather than democratization. I discuss works on democratization only to the extent that they inform my inquiry. For an in-depth review of the literature on democratization, see Geddes (2011). 3 As will become clear in Section 3, I choose not to proceed deductively by identifying a preferred definition of democracy and selecting contributions accordingly (for a taxonomy of such definitions, see Tilly 2007, pp. 7-11). Instead, I proceed inductively by discussing available contributions under the headings chosen by the authors, which include representation, inclusion, and limited government. David Stasavage (unpublished manuscript) discusses these and other instances under the rubric of "early democracy," which, he suggests, was a common form of governance in many parts of the world. Although these developments do not add up to democracy as we know it, they are also clearly not absolutist authoritarian rule. The goal of this article is to offer an overview of existing work that speaks to the challenges of building nonauthoritarian governance where it previously did not exist, and to do so in geographic and temporal contexts that differ from those of contemporary liberal nations. alignment between shared normative commitments and institutions for democratic stability (Bowles 2016). In doing so, it also reminds us that important goods that we normally associate with modern liberal democracies-nonauthoritarian governance, stability, and legality-do not necessarily rely on a given set of societal values and do not have a unique institutional manifestation.

DEMOCRATIC STABILITY IN THE MODERN WORLD
The modern literature on democratic stability has devoted a great deal of attention to explaining the relationship between democracy and economic growth. In particular, modernization theorists (Lipset 1959, Huntington 1968) argued that democracy is more sustainable when economic growth enables the creation of a strong middle class. The last two decades have seen a burst of empirical work, spurred by Przeworski & Limongi's (1997;cf. Przeworski et al. 2000) finding that modernization, as measured by wealth, has no impact on a country's probability of democratizing (but contributes to stability). Many turned to disproving the claim (e.g., Boix & Stokes 2003, Epstein et al. 2006. Others began to tackle the sister question: Does democracy cause economic growth? Until the mid-2000s, the prevailing view was that democracy has a negative or no impact on growth (Barro 1997; for an overview of the literature, see Gerring et al. 2005). But in recent years, a series of important studies showed instead that democracy positively affects growth (e.g., Acemoglu et al. 2014). Taken as a whole, the literature has reached no consensus (Knutsen 2012).
In more recent years, the increasing sophistication of quantitative tools and methods has made the question of the causal relationship between democracy and growth somewhat more insular. In addition, as the threat of democratic breakdown began to reach western developed shores, the question lost traction, in part because it says nothing about when, how, and why democracy will collapse in countries that are already affluent. A new wave of literature, cited above, has begun to re-examine the sources of democratic breakdown [on the global democratic recession, see Diamond (2015)]. Other work has focused on identifying the institutional channels through which democracy (or some aspects thereof ) affects growth. Critical variables include institutions protecting property rights, investment, (a degree of ) state autonomy, and checks on predatory behavior (Przeworski & Limongi 1993), but also institutions fostering political stability (Feng 2003) and the accumulation of human capital (Tavares & Wacziarg 2001, Baum & Lake 2003, Doucouliagos & Ulubaşoǧlu 2008. Along similar lines, the success of the interpretive tools of new institutionalism has pushed to the fore the importance of self-enforcing rules (Hardin 1989;Przeworski 1991Przeworski , 2005Ordeshook 1992;Boix 2003;Acemoglu & Robinson 2006;North et al. 2009;Fearon 2011).
Institutions fostering democratic stability usually depend on the existence of a strong state, capable of shaping political, administrative, and economic processes toward stability (Linz & Stepan 1996). Tilly (2007, p. 15ff ) suggests that democracy requires state capacity, understood as the capacity to "supervise democratic decision-making and put its results into practice." Knutsen (2013) focuses on the interaction of growth, democracy, and state capacity. Andersen et al. (2014) equate notion of consensus is closer to Rawls' idea of overlapping consensus. Rawls postulated an overlapping consensus on a political conception of justice that guides free and equal citizens in cooperating fairly over time as the basis for democratic stability and legitimacy. An overlapping consensus comes about when reasonable citizens engaged in peaceful public reasoning willingly embrace the political conception of justice as a basic module, each citizen working out for themselves how the module fits their broader moral, religious, and philosophical comprehensive doctrines [Rawls 1999[Rawls (1971(1993. Along similar lines, the Athenians developed a consensus on legality as a positive (because it aided coordination) and normative (because it expressed a shared view of good government) tool of rule-making. But the Athenian consensus was the product of bloody strife, rather than peaceful public reasoning. state capacity with administrative effectiveness. How a strong state may come about, however, is a matter of debate (on sequencing, see Fukuyama 2005Fukuyama , 2011Fukuyama , 2014contra Carothers 2007). 5 A somewhat parallel, yet important strand of the literature, often seen as originating in Max Weber's 1905 book The Protestant Ethic and the Spirit of Capitalism, focuses on cultural attributes as critical determinants of democratic stability (Almond & Verba 1963, Inglehart 1990, Putnam 1993, Inglehart & Welzel 2005, Diamond 2008. Despite its periodic resurgence, most recently with datasets like the World Values Survey, the cultural hypothesis is best seen as a contribution to the study of institutions writ large ( Jackman & Miller 1996). I come back to this point in Section 3, where I discuss the role of the interaction between culture and institutions in promoting democratic stability in ancient Athens.
In sum, this literature suggests that stable democracies look very much like liberal democracies, whose critical features are economic growth and strong state institutions that can ensure a robust rule of law, free and fair elections, and the protection of individual rights (Dahl 1971;Diamond 2008;Fukuyama 2011Fukuyama , 2014. These accounts prompt the question of whether democratic stability, or forms thereof, can be extracted from the liberal mold.

THE VIEW FROM PREMODERNITY
What can we learn about democratic stability when we cross the threshold of modernity? Until recently, the literature has focused overwhelmingly on the political transitions that occurred in England in 1688, in the United States in 1776, and in France in 1789 (Moore 1966, North et al. 2009). Although these transitions did not yield "democracy" in any modern sense of the term, it is indisputable that something happened at these historical junctures that enabled a significant leap away from the otherwise dominant authoritarian paradigm. The leaps in England and the United States were followed by other leaps in the same direction; the one in France was not.
Several studies focus on the emergence of institutions of limited government. Perhaps the most influential of these is North & Weingast's (1989) analysis of England's Glorious Revolution. The authors identify the Crown's arbitrary and confiscatory power as the driver of constitutional breakdown. To make the Crown's ability to honor its commitments credible, the constitution that followed the Glorious Revolution established self-enforcing institutions of limited government. Stasavage (2002) enriches North & Weingast's account by focusing on the interaction of economic, religious, and political interests. For Stasavage, stability depended not just on establishing divided government but on the processes whereby veto power was delegated and distributed across parties and cross-issue coalitions. Stasavage then compares England with France, where, he suggests, institutions of limited government alone would not have yielded credible commitment because of a lack of durable political coalitions. The case of England has spawned a number of analyses focusing on the role of self-enforcing rules and credible commitments for the emergence and sustainability of limited governments in Venice, Genoa, and the United States (Weingast 1997, Greif & Laitin 2004, González de Lara et al. 2008, Mittal & Weingast 2013). These studies have contributed to, and in many cases benefited from, a robust literature that views representation as a solution to 5 Mainwaring & Bizzarro (2019) survey three outcomes in third-wave democracies: deepening, stagnation, and erosion. The results are sobering, as only a little less than 30% of these transitions led to democratic deepening. Even more sobering, for only eight countries such deepening put them above the threshold for illiberal (Zakaria 1997) or electoral (Diamond 1999) democracies. Four more countries (the Czech Republic, Estonia, Lithuania, and Slovenia) made conspicuous gains after the transition and maintained a high level of democracy. The factors that contributed to deepening include not only greater state capacity and higher economic development, but also more democratic neighbors, and past experiences with democracy. commitment problems (North & Thomas 1973;North 1981North , 1993Levi 1988;Acemoglu et al. 2005;Acemoglu & Robinson 2000, 2006.
If we expand the inquiry further, we encounter a wealth of recent studies that, while taking us perhaps farther afield from "democracy," center on the development of representative institutions at the local and national levels in late medieval and early modern Europe (providing, among other things, some historical and analytical background that helps contextualize the better-known cases). These studies, often predicated on within-country or cross-country panel data, include analyses of the relationships between fiscal centralization and limited government (Dincecco 2009); between political representation and geography (Stasavage 2010), public credit (Stasavage 2011), and guild revolts (Stasavage 2017); between participative institutions of self-governance and city development [Wahl (2019) finds a positive effect, which declines over time]; and between trade and parliamentary representation [Angelucci et al. (2019) identify municipal autonomy triggered by the commercial revolution as a stepping stone toward parliamentary representation]. Overall, this literature reflects modern findings by suggesting a strong relationship between economic growth and self-governing institutions through advances in the state's administrative capacity. 6 Related to this literature are studies of the development of inclusive political institutions. These studies focus on the reasons that compel elites to share power: fear of revolution (Acemoglu & Robinson 2000, 2006, bargains among a divided or heterogeneous elite that make it in the elite's interest to share power (Lizzeri & Persico 2004, North et al. 2009), geographic and technological constraints that make it difficult for the ruler to appropriate revenue (Mayshar et al. 2017), or the need to overcome time inconsistency problems that limit investment (North & Weingast 1989, Fleck & Hanssen 2006). These investigations have recently engendered a further expansion of the inquiry toward the deep past, including ancient Egypt, Mesopotamia, and Athens (Acemoglu & Robinson 2016, Mayshar et al. 2017, Benati et al. 2019, Carugati 2019b. 7 The literature yields a series of important findings. First, it confirms the modern connection between political development, state capacity, and economic growth. But it also suggests that modern states or postindustrial levels of growth are not necessary for stability. Second, it uncovers building blocks of institutional design that are not unique to liberal democracy but are nonetheless critical to stability: most notably, self-enforcing rules, as well as institutions enhancing the credibility of commitments among actors endowed with political, economic, and violence potential. Third, it offers a rich perspective on the question of stability that accounts for the choices that actors made as well as the geographical and historical settings in which they made them. There are, however, two downsides: First, often these studies lack a theory of change, modeling the specific events that precede the emergence of self-governance as shocks, or neglecting them entirely; second, the literature overall pays much greater attention to the emergence of nonauthoritarian governments than to the factors that may have enhanced their stability. To better understand the sources of democratic stability before the modern period, we must turn to ancient history, and particularly to ancient Athens.

THE CASE OF ANCIENT ATHENS (508-322 BCE)
For about 200 years, ancient Athens sustained the world's first large-scale experiment in democratic governance-a form of participatory democracy in which the entire adult male citizen population (approximately 30-60,000 people) was involved in politics. 8 In the same period, Athens also achieved remarkable levels of economic and social development, as reflected in measures of economic growth (Morris 2004(Morris , 2005Ober 2015), health and urbanization (Morris 2004, Lagia 2015, real wages (Scheidel 2010), and inequality (Kron 2011, Morris 1998, Ober 2015. 9 But Athens was not always democratic, stable, and prosperous. At the end of the fifth century BCE, 10 the city-state's political institutions succumbed to the joint pressure of military defeats and the erosion of democratic norms and institutions. In the span of a decade, Athens experienced five constitutional transitions, lost the empire that was the foundation of its economic structure, was bereaved of half of its citizen population, and ultimately devolved into civil war. After the civil war, the Athenians reestablished democracy. The new democracy remained in place for almost a century, and it presided over another period of economic growth and social development. Athens thus provides a remarkable laboratory to study the sources of, and challenges to, democratic stability before the advent of modernity. In what follows, I touch briefly on Athens' democratic beginnings in order to describe the city-state's institutional structure, and then proceed to discuss the causes of the democracy's breakdown and recovery. The section on breakdown unpacks the "shock," tracking the constitutional upheavals that followed the collapse of democracy, with a particular focus on the origins and evolution of a societal consensus on legality. The section on recovery reconstructs the institutional changes that the consensus inspired. [I discuss these processes in greater detail elsewhere (Carugati 2019a, ch. 2).]

Democratic Beginnings
In the archaic period (ca. 800-500), Athens, like many other Greek city-states, was marred by socioeconomic conflict. Constant vying for power among the elite threatened the social order. In addition, new opportunities for enrichment widened the gap between the rich and the poor. To solve these problems, the Athenians, like other Greeks, turned to law and began to write down measures to regulate the power of officials. Profound economic and demographic changes, paired with a number of welfare-enhancing reforms, especially those of Solon (on which more below), strengthened the position of the masses. Taking advantage of these developments, in 508-507, Cleisthenes changed the rules of the political struggle. Instead of relying on an elite coalition to prevail against his competitor Isagoras, Cleisthenes sought the support of the demos. The demos answered the call, besieging Isagoras (and his main supporter, the Spartan king Cleomenes) on the Acropolis.
The Athenian Revolution paved the way for Cleisthenes's reforms, which created the building blocks of Athens' mature democratic structure. This structure included a deliberative body of 6,000 adult male citizens, known as the Assembly; the Assembly's agenda-setting organ-the Council of Five Hundred-whose members served for one year; the Popular Courts, which featured the participation of hundreds of lay citizens in large juries tasked with resolving disputes; and finally, a large number of magistrates (ca. 700 in the mid-fifth century) responsible for the 8 As Ober (2017, ch. 1) has shown, ancient Athens is the world's first experiment in "basic democracy": that is, a system of collective self-rule by an extensive and socially diverse demos legitimately empowered to seek-and capable of achieving-security, prosperity, and nontyranny. 9 Economic growth: aggregate consumption measured at 0.6-0.9% per annum, as compared to Holland's 0.5%; per capita consumption measured at 0.15% per annum, as compared to Holland's 0.2% and Rome's 0.1%. Low inequality emerges from measures of wealth (0.708 Gini), landholding (0.382-0.386 Gini), and income (0.40-0.45 Gini, including slaves and resident foreigners). 10 All dates in this section are BCE unless otherwise specified. administration of the city-state. Pay for service and selection mechanisms ensured widespread representation and participation. Law played an important role in Athens' democracy. As Forsdyke (2018) argues, in the fifth century the Athenians made important strides toward establishing elements of what we may today call a rule of law: namely, legal supremacy and legal equality. Scrutinies at the beginning and end of a magistrate's mandate incentivized proper behavior on the part of officials charged with implementing policy. Similarly, practices like ostracism enabled the masses to remove prominent, office-less leaders who were becoming "more equal than others." This complex structure was sustained and paid for through imperial revenues. The Athenian empire was the product of the consolidation of Athens' power over the group of poleis that had fought the Persians in the years 490 and 480-479, which came together in an alliance originally led by Sparta. After the Persian Wars, the Spartans returned to the Peloponnese, while Athens and the allies met on the island of Delos to hammer out the details of a new alliance to continue fighting the Persians in the Aegean. The empire contributed to Athenian prosperity by providing revenue in the form of both tribute and rents (Ober 2015).

Democratic Breakdown
Toward the end of the fifth century, democracy collapsed. Military pressures played a role. From 431 to 404, Athens was involved in the Peloponnesian War against her arch-enemy Sparta. In the midst of the war (in 415), the Athenian Assembly enthusiastically voted to send a massive military expedition to Sicily. The campaign proved an utter disaster for Athens. But military failures alone cannot fully account for the collapse of democracy. A crisis of legitimacy also contributed to the collapse. The crisis was rooted in an institutional design defect that can be traced back to the democracy's very establishment (F. Carugati, J. Ober, unpublished manuscript). To counterbalance the threats of tyranny and elite factions, the power of the Assembly was unchecked: No other institution existed to limit the demos' legislative authority, and there were no procedures to collect and archive decisions. As a result, the Assembly could not credibly commit to policy. Its decisions were valid only as long as the demos was willing to respect its previous pronouncements. After Sicily, this lack of credible commitment triggered fear of expropriation among the elite and led the demos to question its own ability to make sound decisions, especially under the pressure of conflict (Thucydides 8.1.1-3, Carugati 2019a). After a century of democracy, the Athenians began to consider constitutional change. Hundred (411-410). When democracy collapsed, the supporters of oligarchy put forth a platform that promoted their preferred constitutional option as a sound, rational response to Athens' crisis. An oligarchic government would address the polis' fiscal needs by abolishing pay for magistracies, funneling state resources to the war effort, and limiting the franchise to those who could serve the state "in person and in purse" (Thucydides 8.65.3). Moreover, if the Athenians were to agree to constitutional change, the supporters of oligarchy alleged, the king of Persia would look favorably toward the city ([Aristotle] Constitution of the Athenians, 29.1; Thucydides 8.53.1). However, as the historian Thucydides (8.68.4) put it, establishing an alternative form of government in Athens after a century of democracy was "no light matter."

The Four
To legitimize their government, the oligarchy of the Four Hundred pledged not a radical break with tradition, but rather the restoration of the constitution of the fathers: the patrios politeia. Classical scholars have traditionally dismissed this claim as an empty slogan. But as Shear (2011) has recently shown, the oligarchs' use of patrios politeia enabled them to appropriate Athens' past as a tool to legitimize their claims to power. Accordingly, the Four Hundred began to use institutions, buildings, and traditions that connected their rule with that of Athens' forefathers. The Four Hundred's appeals to patrios politeia as a source of legitimacy, however, soon proved hollow. Unable to maintain order and security under the relentless pressure of the Peloponnesian War, let alone to restore the polis' finances, their rule was characterized by foreign policy fiascoes (Thucydides 8.56), manipulation of Athens' laws and procedures ( Opponents of the Four Hundred were a mixed breed. The Athenian army, stationed on the island of Samos, was anxious to reestablish democracy. A moderate faction within the oligarchic regime was eager to establish a broad-based government of 5,000 citizens. When the leader of the moderate oligarchic faction, Theramenes, learned that the army on Samos would also support the establishment of a government of 5,000 citizens, he spearheaded the effort to remove the Four Hundred from power (Thucydides 8.92.9-11). Thousand (410-409). The Five Thousand also employed the notion of patrios politeia to legitimize their rule and appropriated buildings and traditions associated with Athens' archaic lawgivers (Shear 2011, pp. 51-60). But unlike their predecessors, the Five Thousand did not engage in extraconstitutional action, did not perpetrate arbitrary acts of violence, and did not suffer military defeats. They actually achieved an important victory in the naval battle of Cynossema.

The Five
The reason for their rapid demise is shrouded in the paucity of our evidence. Previous scholarship identified the Five Thousand as an intermediate regime, created with a view to reestablish democratic rule. But there is no reason to doubt that the Athenians wanted to establish a broadbased oligarchy when they did so. However, because it was born as a compromise government for the purpose of removing the Four Hundred (Thucydides 8.76.6, 8.89.3), the Five Thousand may have lacked critical support once in power.

The first restored democracy (409-404).
When the Five Thousand also collapsed after 11 months, the Athenians restored democracy. Like the two oligarchic governments that preceded it, the new democracy appropriated patrios politeia for its own ends and began to refashion Athens as a democratic city. But what would a new democracy look like, after the collapse of the fifth-century structure? The analysis above suggested that the primary obstacle to democratic stability in the aftermath of Sicily was the lack of credible commitment on the part of the Assembly, particularly with regard to the possibility of going after elite assets to finance the war. To address this problem, the first restored democracy passed a series of important reforms in the areas of fiscal policy and legal institutions.
The Athenians sought to regulate elite disbursements through a new measure that allowed the wealthy to pool resources to discharge an onerous tax: the trierarchy (equipping and commanding a warship for a year). As Gabrielsen (1994, p. 173) suggests, the new system "was the first known major concession made to the propertied class." Through this measure, the demos signaled to the elite that it was willing to find a mutually agreeable way to pay the bills. Expropriation, in other words, would not be forthcoming. But why would the elite trust a demos that could change its mind on a whim? To impose limits on the decision-making power of the Assembly, the Athenians undertook a program of collection, revision, and publication of the laws of the city-a program notably referred to as the revision of the laws of Solon (Lysias 30.2). The revision of the laws suggests a willingness on the part of the demos to tie its own hands by gathering and prominently displaying the rules in the central marketplace and political hub of the city, the Agora. Importantly, the publication of the laws was also a powerful signal against violations such as the ones perpetrated by the Four Hundred.
In appropriating patrios politeia for their own ends, then, the new democracy also appropriated the connection, originally forged by the Four Hundred, between patrios politeia and legitimacy. But www.annualreviews.org • Democratic Stability: A Long View to present themselves as a legitimate government, the demos did not merely appropriate buildings and traditions. In collecting, revising, and prominently displaying the laws of the city, and attributing them to Athens' early lawgiver Solon, the first restored democracy established a connection between the patrios politeia (which they were avowedly re-establishing), the legitimate nature of the government, and the fundamental concept on which such legitimacy was grounded: the concept of legality.
The first restored democracy lasted until 404, when Athens' defeat in the Peloponnesian War led to the imposition of the Spartan-friendly oligarchy of the Thirty. (404-403). Although established by the victorious Spartans, the oligarchy of the Thirty still had to present itself as a legitimate government in order to stay in power. To do so, the oligarchs appropriated the notion of patrios politeia, as previous governments had done. Like the Four Hundred, the Thirty appropriated buildings and traditions (Shear 2011, pp. 170-80). But after the first restored democracy, a patrios government was, at a minimum, a government able and willing to respect the laws of the city. Accordingly, the Thirty turned to their own revision of the laws to eliminate those statutes that threatened their position in power, including some of the laws of Solon ([Aristotle] Constitution of the Athenians, 35.2).

The Thirty
As the Thirty began to employ the laws to remove their political opponents, and to persecute, kill, and expropriate men of means, citizens as well as noncitizens, the Athenians took up arms. When the democratic leader Thrasybulus seized the stronghold of Phyle in northwestern Attica, an increasing number of exiles joined him (Xenophon Hellenica,2.4.5,10,25). Epigraphic and literary evidence reveals that Thrasybulus' army included a diverse group of Athenian citizens, resident aliens, and slaves (Aeschines 3.187). 11 From Phyle, these men marched on Athens to fight the Thirty and restore the patrios politeia (Diodorus 14.32.6). The civil war was brief, but bloody. The major battle, the battle of Mounichia, was fought primarily between the democrats under Thrasybulus and the supporters of the Thirty. The democrats won (Xenophon Hellenica, 2.4.11; [Aristotle] Constitution of the Athenians, 38.1; Diodorus 14.2-3; cf. Carugati 2019a).

Crisis and consensus.
After almost a century of democracy, Athens' institutions collapsed due to a combination of military pressures and a crisis of legitimacy. The Athenians began to experiment with alternative constitutional options. Throughout the following decade, supporters of oligarchy and of democracy employed the notion of patrios politeia to legitimize constitutional change. As the Athenians responded to the excesses of subsequent governments, the meaning of this notion evolved. Under the Four Hundred, patrios politeia may have merely expressed a vague connection with Athens' past. But, by the time of the first restoration of democracy, it became imbued with the concept of legality-a concept germane to the notion of patrios politeia as the constitution of Athens' early lawgivers. Among Athens' lawgivers, one stood out: Solon. The first restored democracy referred to the laws being revised as the laws of Solon; the Thirty's attack on the city's laws, at least according to the Constitution of the Athenians, singled out Solon's laws. The decree that ushered in the new democracy in 403 also referred explicitly to the laws of Solon as the pillar of the new constitutional structure (Andocides 1.83). Solon's archaic legislation emphasized the importance of secure personal and property rights, and identified the city's laws and courts as the institutions tasked with protecting and enforcing such rights. The figure of Solon was also not associated with radical forms of oligarchy or democracy. In his poetry, Solon speaks of himself as an impartial arbiter that stood among factions "as a wolf among dogs"-someone who seeks common ground among competing demands ([Aristotle] Constitution of the Athenians, 4). As we will see in Section 4.3, these principles inspired the institutions of the new democracy.
If we neglect the contentious elaboration of the consensus on legality, it is difficult to understand where a series of new institutions, some of which had no parallels in Greece, came from. If we neglect the elaboration of the consensus, it is also difficult to understand why the victors of the civil war-the democrats-created institutions to tie their own hands.

Democratic Recovery
In 403, Athens hit rock bottom. The defeat in the Peloponnesian War in 404 had left the city without walls, ships, and allies. Athenian finances were at an all-time low, and about half of the male population was dead. For the first time since the establishment of democracy in the late sixth century, the worst of all evils-civil war-had ravaged the city. After the violence perpetrated by the Thirty, revenge was an easy path. The Athenians, however, did not take that path. When arms were laid down, the warring parties ratified an amnesty and drafted a series of institutional reforms. The reforms included the creation of a written law code, an additional legislative institution, new procedures to regulate the process of lawmaking, and measures to define and enforce personal and property rights after the civil conflict. These changes reflect a response to the failures of past governments and were inspired by the consensus on legality elaborated during the constitutional struggles. Taken as a whole, the reforms created a self-enforcing constitutional structure that imposed limits on the previously unchecked power of the Assembly, created incentives for all actors to abide by the new rules, and established institutional mechanisms to punish violators.

Limits.
The reforms imposed limits on the previously unrestrained decision-making power of the demos in the Assembly by introducing another legislative institution, the nomothesia (literally "lawmaking"), 12 and by specifying a series of procedures to be followed in the process of legislation. Moreover, a complex system of checks and balances was created to coordinate the legislative process and define the relative spheres of influence of the two institutions. First, the Assembly maintained the power to pass decrees (psephismata), subject to the provision that decrees could not contradict existing laws (nomoi). Second, laws were the domain of the nomothetai, but their power to pass legislation was in turn limited by the provision that the nomothetai could only be convened by the Assembly. Finally, both decrees of the Assembly and laws of the nomothetai had to conform to the body of existing laws, which were collected and published between 410 and 399.
In addition to limiting the power of the demos, in the aftermath of the civil conflict, the Athenians had to deal with the threat of recurring violence. Pacifying Athens required establishing robust institutional channels to defuse the hostility that prevailed among citizens who had suffered violence, expropriation, exile, and disenfranchisement at the hands of the Thirty and during the civil conflict. For this purpose, the amnesty agreement included a right-of-return clause whereby Athenian citizens (i.e., Athenians who were citizens before 403) could reclaim political and legal rights, and whereby both citizens and noncitizens could reclaim their property. To protect the legal system against a wave of disputes that threatened to crush it, the Athenians defined a series of procedures to determine which claims could go to court (Carawan 2013).
In sum, whereas in the fifth century the Assembly monopolized the decision-making power, in the aftermath of the constitutional struggles legislation became the product of a complex and highly regulated system of checks and balances.

Incentives.
The reforms created incentives for the parties to abide by the rules and refrain from violation by making both supporters of oligarchy and supporters of democracy better off than they would have been under alternative constitutional options. What were these options?
Answering this question requires a measure of counterfactual speculation. Because the supporters of democracy won the civil war, oligarchy was no longer a viable constitutional option in Athens. But various forms of democracy could have emerged. In particular, the democrats could have chosen two alternative paths. First, they could have restored the fifth-century democracy, imposing once again no checks on the people's power. This choice would have likely alienated the supporters of oligarchy, who would have then faced three potential responses: to acquiesce, to resume fighting, or to relocate elsewhere in Greece. While fight and flight implied conspicuous costs, acquiescence was likely to prove unsustainable in the long run. Second, the democrats could have turned against the supporters of oligarchy more openly, for example by choosing the path of retribution and expropriation. This path would have increased the risk that the oligarchs would choose to fight or flee. Compared to these alternatives, then, the new reforms made the supporters of oligarchy better off. Indeed, the new constitution provided conspicuous protections to the vanquished.
But were the democrats better off? The new constitutional structure made the democrats better off than they would have been under the counterfactual options outlined above for three reasons. First, the consensus on legality embodied a solution that would enable the democrats to maintain their preferred constitutional structure-that is, democracy-while limiting its excesses. Second, reneging on this solution meant alienating a section of the population, thus sowing the seeds of renewed civil conflict. Third, given the challenges that the city faced after the civil war, protracted civil conflict was not merely a costly option but a threat to the very existence of the polis. In 403, the financial crisis that began in the first phase of the Peloponnesian War and worsened in the aftermath of Sicily had never been so dire. The emergency fund that the fifth-century democratic leader Pericles had put aside for rainy days was long gone. So were other forms of stored wealth, e.g., golden statues that were melted down during the second phase of the Peloponnesian War (from 412 to 404). The defeat at the hands of Sparta in 404 meant the loss of the city's walls, ships, and allies. Adding insult to injury, the Thirty had compromised the infrastructure of Athens' military and commercial hub-the harbor of Piraeus. The Laurion silver mines, a conspicuous source of revenues in the fifth century, were shut down and their slave workers gone. The countryside lay idle after a decade of Spartan occupation. Since the beginning of the Peloponnesian War, Athens had suffered a plague, 27 almost consecutive years of war, and a civil war. The casualties had decimated Athens' adult male citizen population. And the losses compounded each other: without men (and slaves), Laurion and Piraeus were useless; without Laurion and Piraeus, resources were scarce. Under these conditions, the possibility that Athens' strong neighbors-not only Sparta but also Thebes and Corinth-would take advantage of Athens' divisions to enslave or annihilate the polis was not beyond the realm of possibility (Xenophon Hellenica, 2.2.19).

Enforcement.
What kind of institutional mechanisms protected the new constitutional structure from violation? As I mentioned above, the new constitutional structure revolved around the principle of coordination between the Assembly and the nomothetai. Coordination could be preserved only by policing the integrity of the corpus of laws, to which both Assembly decrees and nomothetai laws had to conform. If the Assembly or the nomothetai passed new decrees and laws that contradicted existing statutes, then the institutional balance of power would crumble, allowing the proposers of contradicting measures to elevate themselves above the laws. Such behavior, associated with the experience of oligarchy, was among the threats that the Athenians sought to inhibit.
The stability of the constitution rested on the mandate bestowed equally on every adult male citizen to police the corpus of the city's laws. Through the procedure known as graphe paranomon, any participant in the course of any given assembly could indict a proposed measure as against the laws (paranomon) or inconvenient (asymphoron) in the sense of failing to foster the interests of the Athenian demos. But this was a risky move. First, the procedure technically allowed "whoever wished" (ho boulomenos) of the roughly 6,000 Athenians sitting in the Assembly to bring a public action against the proposer of a new legislative measure. Second, the grounds for indictment were rather loose. Although one had to show that the new proposal contradicted an existing statute, the vague substantive nature of Athens' laws left ample room to articulate the inconvenience or unsuitability of new proposals. Third, these types of public actions slowed down the legislative process by transferring the matter from the Assembly to the people's court. Already in the fifth century, in the famous case of Arginusae, the graphe paranomon had proved futile for the purpose of policing the laws against the will of an enraged demos (Thucydides 3.36-49; Lanni & Vermeule 2013). At a time when the city was divided and personal resentments were heightened by the recent experience of civil war, the chance that people would use the graphe paranomon to prosecute proposers, rather than proposals, was extremely high. Because indictments could be brought against just about any proposal (including various highly time-sensitive policies, such as military actions), had the Athenians frequently used the procedure to pursue personal feuds, the legislative process would have ground to a halt. Systematic abuse would have jeopardized not only the peace process but also the entire constitutional structure.
To prevent abuses, the Athenians relied on institutional design. The use of the graphe paranomon was regulated through a system of fines and other forms of punishment for frivolous litigation. As in other public cases, punishment awaited a prosecutor who failed to gather one fifth of the votes. The proposer of new legislation was also fined if the graphe paranomon was successful (but only for a year). The new constitution also played an important role because it raised the stakes for would-be abusers by conjuring up the threat of civil conflict.

CONCLUSION
What are the sources of democratic stability? The evidence from the three modern waves of democracy suggests that stability requires robust economic growth, strong states, and liberal institutions. But can we hope to secure democratic stability beyond liberalism? This question is relevant to those developing countries that have little hope of, and perhaps little interest in, embracing the liberal democratic paradigm. But this question is also increasingly relevant to contemporary liberal democracies where the achievements of the twentieth-century liberal order are eroding.
In this article, I summarized the evidence for democratic stability before liberalism. The premodern period yields a body of evidence that sheds light on actors' choices as well as the geographical and historical factors driving political development. Some of this evidence confirms the findings of the scholarship on modern democratic stability. Other evidence, instead, calls attention to elements of institutional design that are independent of the particular cultural and institutional forms of modern liberalism.
When we take a further step back in time, we can draw from the experience of a welldocumented case of large-scale democracy: ancient Athens. Tracking the collapse and recovery of democracy in Athens, I highlighted two conditions that enabled the Athenians to survive the crisis of democracy and strengthen, rather than surrender, their institutions for collective selfgovernance: first, the elaboration of a consensus on legality; second, the ability to build upon this consensus a set of robust, self-enforcing institutions that addressed the causes of previous governments' instability; created incentives for the parties to choose institutional channels to resolve disputes, instead of resorting to retaliation and violence; and provided mechanisms for punishing violators. Notably, the consensus on legality did not require a collective agreement on every aspect of the organization of society, and the institutions built on it channeled rather than eliminated disagreement in the democracy (Carugati 2019a;cf. Farrell & Schneier 2019versus Grzymala-Busse 2019. The case of Athens is significant in two main ways. First, it highlights the importance of alignment between shared norms and appropriately designed institutions for stability. Second, it provides a singular body of empirical evidence to reconstruct how a stability-enhancing consensus among citizens holding different views of the good society might arise in the absence of full adherence to the package of liberal values and institutions. For example, the case of Athens stresses the role of legality for the stability of democracy. But the Athenians never established modern rule-of-law institutions-such as independent judiciaries, strict rules of precedent, or a professional police force. And yet, Athenian legality measures up well against thick modern definitions of rule of law (e.g., Fuller 1964; cf. Carugati 2019a).
The case of ancient Athens has long captured the attention of students of politics, but only recently has the ancient city-state become a case study of institutional development in its own right. This trend must continue. As more interdisciplinary work continues to emerge, Athens and other Greek city-states can shed new light on questions of democracy, democratization, and authoritarianism, as well as questions of economic and social development. Moreover, if early democracy was a common form of government everywhere in the world for much of human history (D. Stasavage, unpublished manuscript), then we must pay attention to this emerging body of comparative evidence. As we learn more about the conditions that enabled humans to sustain stable, nonauthoritarian societies throughout history, we will become better equipped to reverse the forces that endanger our modern democracies.

DISCLOSURE STATEMENT
The author is not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.