The Politics of Takings Clauses

A long-standing consensus exists that the arbitrary or excessive expropriation of private property by a country hurts its economic growth. Constitutions can potentially play an important role in protecting private property from arbitrary or excessive expropriation by offering a range of mechanisms that make it more difficult for governments to renege on the constitutions’ promises. Despite their potential importance, remarkably little is known about how constitutions actually restrict the power of eminent domain and whether such restrictions are associated with reduced de facto expropriation risks. This Essay fills that gap by presenting original data on the procedural and substantive protections in constitutional takings clauses from 1946 to 2013. Its main finding is that no observable relationship exists between de jure constitutional restrictions on the power of eminent domain and de facto expropriation risks. This Essay explores two possible explanations for why constitutional restrictions on the power of eminent domain fail to make a difference in practice. The first is that countries adopt disingenuous promises to bolster their international reputation or to attract foreign aid. The second explanation holds that, although takings clauses have the potential to make society as a whole better off, under some circumstances the benefits of secure property rights fall disproportionally upon economic elites, and a majority of citizens benefit more from expropriation than from increased economic growth. As a result of this dynamic, disagreements over the desired level of expropriation might be built into the constitution’s design. This Essay finds empirical support for the second explanation. Specifically, it finds that real-world constitutional property regimes are often riddled with ambiguities. That is, constitutions often include strong procedural and substantive restrictions on the power of eminent domain but also include “fine print” that can undermine those restrictions, such as provisions that offer a narrow definition of private property, policies that contemplate land reform, provisions that restrict the property rights of foreigners, and clauses that make private property subordinate to the common good. What is more, this Essay finds that when accounting for such fine print, constitutional restrictions on the power of eminent domain do appear to be correlated with reduced expropriation risks. This finding suggests that the effectiveness of takings clauses might depend to a large extent on the politics surrounding their adoption.

property because, considering the risk of expropriation, they discount its future value 4 . Indeed, economists have long emphasized the importance of secure private property rights as a crucial ingredient for economic growth 5 . In fact, one recent empirical study suggests protecting private property from government intervention is the single most important institutional predictor of economic growth 6 . Constitutional scholars have often suggested that constitutions can play an important role in protecting private property from arbitrary or excessive government expropriation 7 . Constitutions contain various mechanisms that can restrict a government's ability to deviate from the constitutions' promises ex post. As a result, constitutions allow governments to credibly commit to respect private property and to signal to investors, capital markets, and individual citizens alike that their property is secure from expropriation. (citation omitted)). 5 See supra note 1. 6 Daron Acemoglu & Simon Johnson, Unbundling Institutions, 113 J. POL. ECON. 949, 953 (2005). 7 North & Weingast, supra note 1, at 805-08 (suggesting that credible constitutional commitments to protect private property allow governments to access capital); see also Daniel A. Farber, Rights as Signals, 31 J. LEGAL STUD. 83, 88-89 (2002) (suggesting that constitutional commitments to respect private property can attract foreign direct investors); John Ferejohn & Lawrence Sager, Commitment and Constitutionalism, 81 TEX. L. REV. 1929REV. , 1929REV. (2003 (describing constitutions as pre-commitment devices and noting that "[a] government that is constitutionally barred from expropriating property is thereby better able to attract capital"); David S. Law, Globalization and the Future of Constitutional Rights, 102 NW. U. L. REV. 1277, 1309-10 (2008) (suggesting that countries can use their bills of rights, and especially their property rights protections, to compete for foreign investment).
Although constitutions can assist states in reaping the long-term economic benefits associated with secure property rights, remarkably little is known of how constitutions actually curb the power of eminent domain 8 or whether such restrictions are associated with reduced de facto expropriation risks. This Essay fills that gap by presenting original and fine-grained data on the procedural and substantive restrictions that the world's constitutions have placed on the power of eminent domain from 1946 to 2013 9 . With this data, this Essay documents cross-country variation in takings clauses, including their development over time. It also provides an empirical exploration of the relationship between de jure constitutional restrictions on the power of eminent domain and de facto or real-world expropriation risks. Its core finding is that no observable relationship exists between de jure takings clauses and de facto government respect for private property. However, when accounting for the ambiguities built into the constitution's design, stronger de jure restrictions on the power of eminent domain do appear to be correlated with reduced de facto expropriation risks. This Essay begins by articulating the conventional wisdom on why takings clauses are supposed to reduce de facto expropriation risks 10 . It starts from the assumption that, for most states, the long-term economic benefits of restricting the power of eminent domain will outweigh the short-term gains of expropriation. Yet, a commitment to curb arbitrary or excessive expropriation will pay off only when it is perceived as credible by investors, capital markets, and private citizens. One way to make this commitment credible is to entrench it in a constitution that comes with a set of mechanisms that make it harder to renege on this commitment in the future. Constitutions make it harder to renege on commitments because they are difficult to amend, can be enforced by the judiciary, and generally come with a set of coordination benefits that make it hard to ignore the document altogether 11 . In sum, the conventional wis- 8 The most comprehensive (qualitative) comparative survey on takings clauses in a set of constitutions is provided by A.J. VAN DER WALT, CONSTITUTIONAL PROPERTY CLAUSES (1999). 9 See infra Part II.A (introducing the data). 10 See infra Part I. 11 See infra Part I.A. dom purports to explain both why states adopt protective takings clauses and why they comply with them in practice. This Essay next suggests two explanations for why the conventional wisdom might not hold and why takings clauses might fail to constrain in practice. First, countries could enshrine disingenuous promises in their constitution to reap international benefits, such as improving their international reputation or attracting foreign aid 12 . They may even attempt to attract investors based on such false promises. If so, we should expect takings clauses to fail when adopted in the absence of genuine constraints and, specifically, when there is no independent judiciary that can enforce the constitution's restrictions on the power of eminent domain. The empirical analysis presented in this Essay lends no support for the thesis that the effectiveness of takings clauses depends on the presence of an independent judiciary. A second reason for why takings clauses might fail to constrain in practice is that a majority of citizens might believe that some degree of expropriation benefits them more than a strong protection of private property 13 . When this is the case, the constitutional commitment to protect private property is not necessarily disingenuous but might be surrounded by substantial societal disagreements. How such societal disagreements will affect the takings clause's future operation depends on the politics surrounding its adoption. When those who favor some degree of expropriation control the constitution-making process, it is likely that the constitution will simply enshrine fewer restrictions on the power of eminent domain. By contrast, when economic elites that benefit disproportionally from secure property rights control the constitution-making process, these elites will adopt protective takings clauses that might subsequently be undermined by future democratic politics 14 . A last possibility, explored at some length in this Essay, is that neither economic elites nor popular majorities control the process entirely. As a result, the competing interests of different societal groups will be built into the constitution's design. Indeed, this Essay shows that real-world constitutional property regimes are riddled with ambiguities. Specifically, some constitutions not only include a number of procedural 12 See infra Part I.B.1. 13 See infra Part I.B.2. 14 See infra Part I.B.2.

The Politics of Takings Clauses
and substantive restrictions on the power of eminent domain in their takings clause, but also include "fine print" that is typically separated from the main takings clause but that nonetheless could be used to circumvent the guarantees in the main takings clause. Examples of "fine print" include provisions that offer a narrow definition of private property, policies that contemplate land reform, provisions that restrict the property rights of foreigners, and clauses that make private property subordinate to the common good.
This Essay finds that when accounting for takings clauses' "fine print," that is, the ambiguities that are built into constitution's design, constitutional restrictions on the power of eminent domain do appear to be correlated with reduced de facto expropriation risks. In other words, when the constitutional commitment to restrict the power of eminent domain is unambiguous, stronger restrictions on the eminent domain power are correlated with reduced expropriation risks. This finding suggests that the effectiveness of constitutional takings clauses is to an important extent determined by the politics surrounding their adoption.
This Essay unfolds as follows. Part I articulates the conventional wisdom on why constitutional takings clauses matter as well as two reasons for why this wisdom might not hold in practice. Part II introduces the dataset on takings clauses and documents cross-national variation in takings clauses, including how they have developed over time. Part III contrasts the de jure takings clauses with data on de facto expropriation risks. Part IV explores whether takings clauses matter under some circumstances only. Following the theories developed in Part I, Part IV primarily explores "false positives," that is, states that adopt takings clauses either for entirely disingenuous reasons or in the face of substantial societal disagreements over the desired level of expropriation 15 . It also provides a preliminary exploration of "false negatives," that is, countries that respect private property de jure but do not restrict the power of eminent 15  domain de facto 16 . The core empirical finding from Parts III and IV is that takings clauses appear to have no impact in the aggregate but that they do appear to matter when the constitution unambiguously commits to protect private property and omits internal contradictions. Part IV concludes by describing the limitations of the findings presented in this Essay and suggests future directions for the comparative constitutional law literature.

A. The Conventional Wisdom
It is widely accepted that protecting private property from arbitrary or excessive expropriation contributes to long-term economic growth 17 . When private property is secure from expropriation, citizens make more productive use of their property, governments can more readily access capital on capital markets, and investors are more likely to invest in the country 18 . Consequently, for most societies, the long-term economic benefits that are associated with secure property rights will far outweigh the short-term gains of expropriation.
Yet, the benefits associated with restrictions on the power of eminent domain will materialize only when the restrictions against arbitrary or excessive expropriation are credible 19 . Capital markets, investors, and private citizens are likely well aware of a potential time-inconsistency problem that plagues government commitments to respect private property: though governments appreciate the benefits of secure property rights in theory, they might be tempted to renege on these promises once they are put in place 20 . Expropriation will be appealing when the state is short 50 Seqüência (Florianópolis), n. 71, p. 43-94, dez. 2015 The Politics of Takings Clauses on capital and the survival of the regime is at stake 21 . In such times, politicians whose tenures depend on regular elections might discount the future value of property protection and attempt to reap the short-term benefits of expropriation even at the expense of long-term economic growth. Because of these incentives, economic actors will invest only when they trust that their investments will not be undermined by future government action 22 .
Constitutions have long been regarded as devices that can make commitments credible 23 . One of the fundamental goals of constitutionmaking is to solve the time-inconsistency problem that plagues many government commitments. Constitutions are supposed to be the ropes that, in Homer's parable, bind Ulysses to the mast so that he can resist the singing of the Sirens 24 . Constitutional theorists have proposed various mechanisms that make it harder for a government to renege on its promises and thus make constitutional commitments credible 25 . First, the constitution is the supreme authority of the legal system and is typically harder to amend than ordinary legislation 26 . Second, the constitution's rules and principles are commonly guarded by a judicial body equipped 21 Id. at 807 (noting that the incentives to renege on commitments to protect private property ex post are articularly large in times of crises, such as warfare, because at such associated with constitutional reform). 22 For example, as Professors North and Weingast showed in their study on seventeenthassociated with secure property rights only once it made its commitments credible by appointing Parliament as the key bearer of the crown's treasure chest. Id. at 804. 23  with the power to strike down laws and regulations that contradict these rules and principles 27 . Third, the constitution helps a state to coordinate upon a set of conventions by which it governs itself and that become harder to change with time 28 . The very act of writing a constitution sets conventions that clarify the rules by which different actors must play the political game. For example, the constitution stipulates how many branches of government there are, how they are elected and composed, and what they can and cannot do with their power. Even though political actors might not value a given constitutional rule, they do value the general notion of having rules, which facilitate better government coordination. Because having conventions is valuable in and of itself, it becomes harder to ignore the parts of the constitution that are perceived as less desirable because doing so would undermine the coordination benefits of the constitution as a whole 29 . In sum, the constitution provides a range of mechanisms (such as entrenchment, judicial review, and coordination) that make it harder to deviate from those promises ex post, thereby serving as the metaphorical ropes that tie a nation to its commitments 30  like Ulysses's ropes, however, the constitutional protection mechanisms are not fail-safe. A constitution can ultimately be amended, courts can be ignored, and the rules of the game can be recoordinated 31 . Each of these, however, imposes substantial costs. Thus, even though a constitution does not make it impossible for a government to renege on its constitutional promises, it does raise the costs of doing so. The costs of reneging on constitutional commitments are likely lower when these commitments are counter majoritarian in nature. That is, when constitutional commitments serve to protect minority interests against the majority, majorities can fairly easily amend the constitution, appoint sympathetic judges to the court, or use their political clout in some other way to change the rules of the political game. Indeed, these scenarios were precisely James Madison's concern when he described the bill of rights as nothing but "parchment barriers" 32 . According to Madison, constitutions can prevent the problem of "faction," that is, minorities taking advantage of the majority 33 . But Madison was more pessimistic about constitutions' abilities to systematically protect minorities from majority rule 34 . He feared that when the majority wants to violate the minority's rights, the metaphorical ropes that tie a state to its earlier commitments would be unbound 35 . Of course, it does not necessarily follow that countermajoritarian constitutional constraints are meaningless altogether. Even countermajoritarian constraints raise the costs of noncompliance 36 . As Professor Hardin argues, people acquiesce even to unpopular constitutional rules because they value having a set of 31 See id., at 682-83 (suggesting that many social science theories on precommitment do not actually explain why constitutional constraints endure in the future). 32 THE FEDERALIST NO. 48, at 221 (James Madison) (Jim Manis ed., 2014). 33 Id. NO. 10, at 43-44 (James Madison). 34 Levinson, supra note 28, at 667 (noting that the bill of rights as originally conceived was "meant not to protect against majoritarian tyranny . . . but, quite the opposite, to bolster 35 RIGHTS 160, 161 (Jack N. Rakove ed., 1998) [hereinafter Madison Letter] (noting that most needed" and "[r]epeated violations of these parchment barriers have been committed by overbearing majorities in every State"). 36 For an extended discussion, see Levinson, supra note 28. conventions to conduct their political affairs, and because coordinating upon an alternative set of rules is costly 37 . In general, however, it will be more difficult to renege on constitutional commitments that enjoy majority support. Although those in power might at times be tempted to subvert the majority's wishes, if these wishes are enshrined in the constitution, then a number of constitutional mechanisms might make it near-impossible to do so in fact. Courts are likely staffed with judges who support the constitution's protections, and court decisions can act as "fire alarms" that set off political mobilization 38 . Moreover, where constitutional commitments are majoritarian in nature, proposed constitutional amendments will be voted down, and attempts to undermine the constitutional order will induce political opposition 39 . Thus, constitutional constraints are at their strongest when they enjoy popular support.
Whether constitutional takings clauses are majoritarian in nature, and thus act as strong constraints on the government, is a difficult question. On the one hand, takings clauses could be seen as majoritarian because they contribute to long-term economic growth, which benefits the nation as a whole. Indeed, economists have shown that a lack of secure property rights substantially hampers a country's economic development 40 . Conversely, constitutional restrictions on the power of eminent domain improve economic growth and therefore can make the country as a whole better off 41 . One way to view property clauses, therefore, is as 37 change, and citing the potentially destabilizing nature of the American Electoral College, which remains unchanged after the 2000 election). Madison, by contrast, ultimately came to believe that countermajoritarian rights could work if the structural part of the supra note 28, at 668. 38 39 See Madison Letter, supra note 35, at 162 (describing rights "as a standard for trying the validity of public acts, and a signal for rousing [and] uniting the superior force of the community"). 40 See supra note 1. 41 Id.
The Politics of Takings Clauses majoritarian in nature. On this view, we would expect that constitutional protections against expropriation are particularly effective in constraining future government action. On the other hand, viewing takings clauses as majoritarian is also problematic because, in most societies, property is unequally distributed, and a majority of people might believe that they benefit more from expropriation than from the increased economic growth associated with secure property rights. In that case, constitutional protections against expropriation are likely to be undermined by democratic politics and will fail in practice. The next Part will explore in more detail how the countermajoritarian characteristics of takings clauses might undermine their operation.

B. Revisiting the Conventional Wisdom: Disingenuous and Halfhearted Promises
There are at least two scenarios under which takings clauses fail to protect private property in practice. First, states might adopt takings clauses in bad faith, making the procedural and substantive restrictions ostensibly placed on the power of eminent domain mere deceptions. Second, states might adopt takings clauses when they are only partially committed to protect private property because different groups stand divided over the extent to which the power of eminent domain ought to be curbed. In both scenarios, the conventional wisdom would likely not hold, and the level of de jure takings protections can be stronger on paper than in practice.
1. Disingenuous Promises. One reason why restrictions on the power of eminent domain could be stronger on paper than in practice is that some countries enshrine disingenuous promises into their constitutions. The constitution of North Korea, for example, states that "[t]he State shall protect private property and guarantee its legal inheritance" 42 . In practice, however, all property belongs to the state and the people enjoy no private property protections at all 43 . Disingenuous promises emerge when regimes want to avoid subjecting themselves to any genuine constitutional constraints. For authoritarian regimes, the benefits of evading constitutional constraints often outweigh the economic benefits of protective takings clauses 44 . Authoritarian rulers, after all, do not necessarily seek to maximize general welfare, but rather seek to secure their own grip on power 45 . Such regimes can either avoid making any constitutional promises at all or adopt false constitutional promises 46 . There are a number of reasons why they might choose the latter. First, they could do so simply to bolster their international reputations 47 . A body of sociological literature has demonstrated that the international community is increasingly characterized by standardized scripts of statehood, including standardized constitutional templates that protect democracy, liberal values, and property rights 48 . A formal constitution that includes the values and principles enshrined in this standardized script has become a sign of modernity and a ticket for entry into "world society" 49 . By adopting such a constitution, states signal conformity to the values of world society, and can improve their standing reasons-countries-fall-apart/ [http://perma.cc/832S-A8DT] (observing that "North Korea's economic institutions make it almost impossible for people to own property; the state owns everything, including nearly all land and capital"). 44 47 Ginsburg & Simpser, supra note 45, at 6 (noting that constitutions in authoritarian regimes often serve as "billboards" that display an attractive picture to the world). in the international community 50 . Thus, by conforming to international norms against expropriation, states might reap benefits associated with an improved international reputation.
A second reason why states might adopt facade takings clauses is to attract bilateral and multilateral aid flows 51 . Indeed, both foreign aid donors and international financial institutions have often made their assistance conditional upon respect for human rights, including property rights 52 . Aid donors, moreover, have offered technical assistance for ruleof-law reforms that include protective property regimes 53 . In both cases, the involvement of aid donors might prompt drafters, acting in bad faith, to adopt protective takings clauses to reap the benefits offered. Finally, states could feature facade restrictions on the power of eminent domain with the intent of luring foreign investors, only to later expropriate the invested property. Foreign investment is characterized by the so-called obsolescing bargain, which holds that investors are in a dominant position ex ante when they are deciding on where to invest, but that they lose this advantage after they have made their investment 54 . Therefore, if ill-intending states are able to attract investors with false promises, these states will be in a strong position to expropriate them afterwards. Of course, international investors are unlikely to be fooled so easily. Investors devote substantial time and resources to assessing whether their investments are secure, and they are likely to know when governments are not genuinely committed to protecting private property. But where promises are cheap, constitution-makers might consider it worth a try.
False promises can be detected by a conspicuous absence of real enforcement mechanisms 55 . Accepting genuine constraints on the power of eminent domain can be a dangerous strategy for regimes that seek to govern without constraints. When doing so, they risk reducing their grip on power because the constitutional mechanisms that curb eminent domain could be turned against them in other areas as well. For example, when the Sadat regime in Egypt made a conscious effort to attract foreign investment by adopting a strong takings clause and establishing an independent constitutional court to enforce it, the government was later forced to accept court rulings that protected individual rights, 56 leading one commentator to describe the Egyptian constitution as "a joke that turned serious" 57 . For those who want to govern without constraints, the more plausible strategy, therefore, might be to adopt constitutions that promise rights, including property rights, but to not offer any mechanisms to enforce these rights against the government 58 . the "obsolescing bargain," which holds that investors are in a dominant position ex ante, at the time host countries are persuading them to invest, but that they lose this advantage ex post, and will therefore be cautious on when to invest). 55  Perhaps the most important mechanism for enforcing the constitution's restrictions on the power of eminent domain is an independent judiciary with the power of constitutional review 59 . Courts can strike down laws that expropriate excessively, and they can act as a neutral arbiters in property disputes between the government and private parties 60 . Investors have long accepted the notion that an independent judiciary is crucial for the protection of property rights from government expropriation. As one study from the 1970s notes: The presence of a strong, independent, and competent judiciary can be interpreted as an indicator of a low propensity to expropriate. If this judicial system is strong, independent, and competent, it will be less likely to 'rubber stamp' the legality of an expropriation and more likely to accede to a standard of fair compensation. The effect of this would be to lower the propensity of the host nation government to expropriate 61 . constitutions that exist for "cosmetic" purposes as "sham" constitutions, and observing that "[t]he principal function of a sham constitutional text is to deceive"); A.E. Dick This insight suggests that when states adopt constitutional takings clauses in the absence of an independent judiciary that can enforce these clauses, their commitment to protect private property might be disingenuous. Thus, in countries that purport to constitutionally restrict the power of eminent domain, the absence of an independent judiciary is a red flag 62 .

Halfhearted Promises: Faction and Disagreement.
Another possible explanation for why eminent domain restrictions might be stronger on paper than in practice is that the restrictions enjoy only partial popular support. That is, the promise to respect private property is not entirely disingenuous, but it lacks support from important segments of society. Although curbing the power of eminent domain can make society as a whole better off, takings clauses are hardly majoritarian when the benefits of constitutional restrictions on eminent domain fall disproportionally upon economic elites. In general, restrictions on the power of eminent domain disproportionally benefit those who possess more property. Moreover, these same economic elites might also receive a disproportionate share of the increased economic welfare flowing from secure property rights. When wealth is unequally distributed, a majority of citizens might benefit more from expropriation than from the long-term economic benefits associated with secure property rights. For the bottom part of the income earners, expropriation will be appealing when they expect that only a small part of the future gains will come their way, because they discount the future, or both. Regardless of their exact motivations, competing preferences for the desired level of property protection likely affect the operation of the takings clause in practice.
How different preferences for restricting eminent domain affect the takings clause's operation likely depends on the circumstances surrounding its adoption. Where democratic majorities with a preference for some degree of expropriation dominate the constitution-making process, the constitution might simply enshrine fewer restrictions on the power of em-62 sometimes credibly commit to private property, as the Egyptian case suggests. The focus of the empirical analysis in Parts III and IV will therefore be on judicial independence, while controlling for democracy. See also note 133 supra for a further discussion of the possible impact of democracy.
The Politics of Takings Clauses inent domain. To illustrate, the drafters of the 1983 Canadian Charter decided to omit a takings clause altogether for fear of replicating the American Lochner experience 63 . Indeed, there exists substantial cross-country variation in the strength of the restrictions placed on the power of eminent domain 64 . These constitutional differences presumably reflect how different constitution-makers balance the importance of private property against other values. Thus, where a portion of the population favors restricting the power of eminent domain, and this group is influential in the constitution-making process, the constitution will simply enshrine fewer de jure restrictions on the power of eminent domain.
By contrast, when the constitution-making process is dominated by economic elites who value secure property rights more than popular majorities do, this might result in a situation in which restrictions on the power of eminent domain are stronger on paper than in practice. According to a body of literature, constitutions often serve as tools for elites to insulate their policy preferences from future democratic decisionmaking 65 .
Professor Hirschl has found that economic elites who anticipate losing power will constitutionally entrench strong property rights to protect their property from future expropriation by democratic majorities 66 . Likewise, historian Charles Beard has famously argued that the desire to preserve wealth and privilege was the guiding rationale behind the U.S. 63 Sujit Choudhry, The Lochner Era and Comparative Constitutionalism, 2 INT'L J. CONST. L. 1, 3-5 (2004) (suggesting that the Canadians wanted to avoid the American Lochner experience, whereby "the U.S. Supreme Court struck down close to two hundred state and federal laws" that infringed upon economic freedom and private contract). 64 66 See generally HIRSCHL, supra note 53, at 49 (observing that "[e]conomic elites may therefore view the constitutionalization of rights, especially property, mobility, and harmful large-government policies of an encroaching state").

Mila Versteeg
Constitution, which, he says, largely protected the economic interests of the wealthy 67 .
If a constitution disproportionally reflects the interests of economic elites, its restrictions on eminent domain are unlikely to align with majoritarian preferences. As a result, the various mechanisms designed to keep a government to its constitutional promises might break down. Democratic majorities can use their political clout to undermine the takings clause in different ways 68 . To illustrate, in the United States, President Roosevelt's court-packing plan forced the Supreme Court to adhere more closely to democratic preferences on property protection 69 . Likewise, the Indian Supreme Court backed down from protecting property against social welfare legislation after a two-decade long battle with the Indian legislature in which the Indian legislature kept passing constitutional amendments to overturn Supreme Court decisions 70 . Where such ideological shifts take was between the "natural aristocracy" and common people representing "turbulent democracy" and that "the protection of property rights lay at the basis of the new system"). But see FORREST MCDONALD, WE THE PEOPLE: THE ECONOMIC ORIGINS OF THE CONSTITUTION 398 (Transaction Publishers 1992) (1958) (arguing that those "[i]t is therefore not even theoretically possible to devise a single set of alignments on the was the principal motivating force"). 68 Majorities may push for constitutional amendments and reduce de jure takings protections. In that case, however, we should not expect a disconnect between de jure and de facto property protections. 69  2005) (describing President Roosevelt's court packing plan and noting that "this change of direction had a profound impact on the judicial protection of property rights. The Supreme Court now deferred to legislative judgments about economic policy and largely abandoned its historical dedication to the sanctity of private property and contractual arrangements"). 70 For an overview, see VAN DER WALT, supra note 7, at 192-206.
The Politics of Takings Clauses place, the country will either move to a lower level of property protection de jure (through amending the constitution) or offer stronger property protections on paper than in practice.
Another possibility is that neither popular majorities nor elites dominate the constitution-making process entirely and that the competing preferences of different groups are built into the constitution's design. That is, the constitution simultaneously reflects both the values of economic elites that favor strong restrictions on eminent domain and those of popular majorities that want to carve out exceptions to these restrictions. The resulting constitution has built-in contradictions and ambiguities that allow it to be interpreted in different ways, depending on how the political landscape changes after the constitution is ratified.
The use of contradictory provisions, or strategic ambiguities, has been documented in different areas of constitutional law. Professors Bali and Lerner document this phenomenon with respect to religion: Constitution-makers enshrine contradictory religion clauses when different religious groups stand divided over the appropriate role for religion in society 71 .
They note that "[p]articularly in cases where the constitutional status of religious law was an important axis of debate, the outcome documents frequently embraced two approaches often considered to be mutually exclusive in the comparative constitutional literature" 72 . In a recent study, Professor Negretto documents a similar phenomenon for electoral rules in Latin American constitutions 73 . He notes that, as the democratic coalitions that write and revise constitutions have become more inclusive, constitutions have witnessed the adoption of electoral rules that improve the democratic responsiveness of government officials 74 . At the same time, dominant groups within these coalitions have often been able to use their power to expand executive power, resulting in contradictory constitutional designs that simultaneously enhance democratic responsiveness and expand executive power 75 .
In the same vein, Professor Lombardi has suggested that constitutions can incorporate strategic ambiguities to accommodate the competing ideologies of different societal groups in a single document 76 . Lombardi illustrates his claim in the context of the 1971 Egyptian Constitution, which contained religious, libertarian, and authoritarian provisions that were in direct tension with each other and that reflected the diverging values and interests of competing groups. Lombardi observes that the constitution "reflected considerable confusion about the nature of the state and the ideology of its governing institutions," allowing one to imagine the state in radically different ways: Pan-Arabist or Egyptian Nationalist, socialist or capitalist, and secular or Islamic 77 . These ambiguities made it possible for warring factions to reach consensus because each faction believed that the constitution could protect their ideology if they were to hold power in the future 78 . Indeed, the observation that societal disagreements might render constitutions internally inconsistent traces back at least to Carl Schmitt's notion of the dilatorischer Formelkompromiss ("disingenuous compromise"), which produces a constitution that, in Schmitt's words, "satisfies all contradictory demands and leaves, in an ambiguous turn of phrase, the actual points of controversy undecided," and therefore provides nothing but a "semantic jumble of sub- of why constitution-makers often defer decisions on contested issues through the use of "by law" clauses). Florianópolis), n. 71, p. 43-94, dez. 2015 The Politics of Takings Clauses stantively irreconcilable matters" 79 . It is perhaps no surprise, then, that the politics surrounding private property render constitutional property regimes similarly ambiguous. As subsequent parts of this Essay elaborate, constitutions are often riddled with contradictions when it comes to property protections. To illustrate, several constitutions establish a very narrow meaning of "private property" that the nominally robust takings clause protects 80 . Other constitutions enshrine a strong takings clause in the bill of rights, but include a separate section on state policies, which curtails property rights in the name of land reform or urban planning 81 . Still others make sweeping claims that private property should serve a social purpose, or they subject foreign citizens to different property rights regimes 82 . In many cases, these contrary provisions are not enshrined in the bill of rights but in various other parts of the constitution, such as a section on government policies or general principles 83 . And perhaps even more surprisingly, these contrary provisions are found especially in constitutions that place strong restrictions on the power of eminent domain in their main takings clause. In other words, the stronger the protections in the main takings clause, the higher the probability that contrary provisions will appear in other parts of the constitution 84 . These contrary provisions thus act as fine print that should alert the careful reader that different parts of society were unable to agree on the desired level of property protection.

Seqüência (
Where such ambiguities are built into the constitution's design, the constitution sends a mixed message: it both promises not to expropriate while including provisions to undermine the same. How these contrary provisions will affect the operation of the takings clause in practice likely depends on future political coalitions and their preferences. Where majorities prevail, and they benefit insufficiently from the economic benefits associated with secure property rights, it is likely that the fine print provisions will swallow the protections enshrined in the main takings clause. By contrast, where elites govern or majorities do value the long-term economic benefits associated with secure property rights, the restrictions will prevail over the exceptions. Although both are plausible in theory, the empirical findings in the next Part of this Essay demonstrate that in the face of constitutional contradictions, the exceptions become the rule 85 .

II. TAKINGS CLAUSES IN THE WORLD'S CONSTITUTIONS
Takings clauses have become a near-universal ingredient of the world's constitutions. Today, no less than 94% of all constitutions include a takings clause 86 . Constitutional takings clauses do not prohibit expropriation altogether, but rather define and limit the circumstances under which a government can expropriate, thereby reducing the arbitrariness of expropriation. Substantial variation exists in the kind and number of restrictions that are placed on the power of eminent domain. Specifically, takings clauses differ from each other in the amount of compensation they require, when compensation ought to be paid, the circumstances under which expropriation is allowed, and the procedural guarantees that must 84 See infra Part II.B. 85 See infra Part IV.A.2. 86 See infra Part II.A (introducing the data).
The Politics of Takings Clauses be followed. Such differences presumably reflect differences in the composition and preferences of constitution-makers in different countries 87 . As a result of these differences, some constitutions ultimately provide stronger protections against expropriation than others. The remainder of this Part will explore and document this variation.

A. Variation in Takings Clauses
The empirical exploration of constitutional takings clauses is made possible by an original dataset on property protections in national constitutions from 1946 to 2013. This data was collected as part of a larger database on the world's constitutions. Specifically, for each constitution written since 1946, the author hand-coded 237 variables that capture rights, rights-related policies, and their enforcement. The database takes account of the amendment and replacement of constitutions, thus tracking changes in the prevalence of these 237 constitutional features over time. This general dataset contains seventeen variables relating to constitutional property rights, which are used in this paper 88 . The coding of national constitutions, of course, requires a range of decisions on what constitutes a constitution, as well as a range of decisions on how to code each constitution. The coding protocol for the database is documented in greater detail in the author's earlier work 89 .
A potential shortcoming of this data is that it only captures the text of the constitution and excludes judicial interpretations of the text. In theory, some of the substantive and procedural restrictions on the power of eminent domain that help nations credibly commit to respect private property might be found in judicial interpretations, rather than in the text of the constitution itself. Most countries, however, set out their constitu- tional property regime in great detail 90 . One likely motivation for doing so is to limit judicial discretion in interpreting such provisions. Another motivation is that explicit constitutional protections send a stronger and clearer signal to foreign investors that their property is protected 91 . Although investors could potentially consult judicial decisions, these are less visible, less accessible, and more easily overturned. As a result, we may expect that, in most cases, the explicit textual limitations on the eminent domain power are a fairly accurate representation of the constitutional limitations on expropriation. The remainder of this Part will explore the various explicit constitutional restrictions that states have placed on the power of eminent domain.

B. Compensation Requirement
A substantial portion of the world's constitutions require the government to compensate expropriated property. What constitutes the optimal level of compensation for expropriation is the subject of a voluminous literature in law and economics 92 . The basic premise of this literature is that if the required level of compensation is too low, the government will expropriate more than is socially optimal, and if the level of compensation is too high, the government will expropriate less than is socially optimal. The extensive literature notwithstanding, there appears to be little scholarly consensus on the optimal level of compensation 93 . Also the world's constitutions vary in the amount of compensation they demand. 90 Versteeg & Zackin, supra note 26, at 1703 (noting that foreign constitutions are generally more detailed than the U.S. Constitution). 91 See, e.g., Farber, supra note 6, at 88-91. 92

The Politics of Takings Clauses
Some takings clauses do not require any compensation at all, 94 while others require the government to pay compensation but do not specify how much compensation ought to be paid 95 . Takings clauses that do specify the level of compensation roughly fall into two categories: (1) those that explicitly require either "full compensation" or "market value" compensation, and (2) those that use more ambiguous language such as "fair," "adequate," or "equitable" compensation.
The drafting of the 1994 South African interim constitution illustrates the stakes of this choice. In the negotiations over the takings clause, the National Party that had governed the country under apartheid wanted to enshrine a requirement of full market value compensation in the takings clause, to protect the property interests of their wealthy constituents. By contrast, the African National Party, representing the majority of blacks that had suffered repression under the apartheid regime, wanted the takings clause to merely require "equitable" compensation, presumably to ease the redistribution of wealth to their constituents in the future 96 . Figure 1 depicts the prevalence of each of these options. It shows that the majority of constitutions demand compensation, but use ambiguous language on how much compensation is required. As of 2013, 62.7% of all constitutions with a takings clause opt for this approach. The number of constitutions that unambiguously require full or market value compensation is much lower, although they have been increasing in number since the 1990s. Today, 15.3% of all constitutions with a takings clause require full or market value compensation.

C. Timeliness of Compensation
Takings clauses not only differ in how much compensation they require, but also differ in when they require compensation to be paid. Specifically, among the takings clauses that demand compensation, constitution-makers have opted for various approaches: (1) some require compensation to be paid prior to expropriation, (2) some require compensation to be paid promptly or within reasonable time after expropriation, and (3) some do not impose a timeliness requirement at all. Presumably, constitution-makers that add language requiring prior compensation favor stronger protection of property rights than those who require prompt compensation or omit a timeliness requirement altogether. Figure 2 depicts the prevalence of each of these three options among the constitutions that stipulate a compensation requirement. It reveals that most constitutions do not include a timeliness requirement for compensation. Perhaps most strikingly, the percentage of constitutions that require prior compensation has decreased over time, from about 56.5% in 1946 to 31.6% today. By contrast, the number of countries that require prompt compensation has increased since the 1960s, so that, today, 20.3% of all constitutions that include a compensation requirement demand prompt compensation.

D. Due Process Requirement
Takings clauses also vary in whether they include a due process requirement, that is, a requirement that certain procedures are followed in order to expropriate property. Such a due process requirement can take different forms. Some constitutions require that expropriation be done by law 97 , while others state that nobody should be deprived of property "without due process of law," 98 or that expropriation should be done on "just terms" 99 . Figure  prevalence of this requirement has been similar in past decades. Some constitutions not only include a general due process guarantee, but also explicitly demand judicial involvement in expropriation. By doing so, constitution-makers ensure the involvement of a neutral and independent arbiter in disputes between the government and private property holders, thus making it harder for the government to expropriate arbitrarily. Figure 3 reveals that currently 36.7% of all constitutions with a takings clause explicitly require judicial involvement in expropriation, and that this number has steadily increased over time.

E. Public Interest Requirement
Many constitutions that contain a takings clause not only require the government to follow certain procedural requirements, but also place substantive limitations on the conditions under which the government can expropriate by requiring that expropriation serves a public interest. Such substantive restrictions on expropriation arguably make it harder to expropriate. Figure 4 depicts the percentage of countries whose takings clause stipulates a public interest requirement. It reveals that constitutions commonly require expropriation to serve the public interest. Notably, the graph also reveals that the prevalence of public interest guarantees decreased in the 1970s and 1980, but increased again in the 1990s 100 . Today, 85.9% of all constitutions with a takings clause include a public interest requirement.

F. Towards an Expropriation Index
Some takings clauses offer stronger procedural and substantive protections against expropriation than others. To capture the extent to which any given constitution protects against expropriation, the variables discussed above can be combined into a single expropriation index. Specifically, a country receives the value of 0 if it does not have a takings clause or it does not provide any of the substantive or procedural limits on takings discussed above. Countries receive one additional point when: • The takings clause requires compensation. • The takings clause requires "full" or "market value" compensation. • The takings clause requires timely compensation. 100 nationalization movement during this period, while the increase in the 1990s presumably • The takings clause requires prior compensation. • The takings clause contains a due process requirement. • The takings clause requires judicial involvement in expropriation. • The takings clause includes a public interest requirement.
The resulting additive scale is based on the assumption that each of these individual components is desirable from an investor's perspective and that the constitutional adoption of more of the features listed above makes private property more secure from government intervention. At the same time, it does not assume that some provisions serve as a precondition for the fulfillment of other provisions, or that without some of these provisions property rights would be undermined entirely. The resulting additive scale ranges from 0 to 7. It has a scale reliability coefficient of 0.78 ("Cronbach's alpha"), which means that the correlation between this scale and all other possible seven-point scales that could be constructed from the same components is 0.78 101 . performing common factor analysis with an orthogonal rotation, it turns out that one latent factor explains about 79% of the shared variation of the various items in the scale. Both the scree elbow test and the Kaiser test moreover suggest that only one factor should additive scale in the analysis reported below, the results remain similar. The remainder of this Essay will use the additive scale in the primary analysis because it is easier to interpret. See note 124 supra for a discussion of the results when using factor scores instead of the additive scale.

G. Fine Print
Some constitutions enshrine provisions that contradict the procedural and substantive guarantees against expropriation included in the main takings clause. Specifically, some constitutions include "fine print" that is typically separated from the main takings clause, but that nonetheless could be used to circumvent the guarantees therein.
To a careful reader, a number of provisions might stand out as red flags that suggest that constitution-makers carved out exceptions to the main takings clause. These provisions serve as fine print to the main takings clause. Although there is no single way to draft fine print, there are some fairly standard provisions that appear in multiple constitutions and that lend themselves to quantitative coding.
Particularly suspicious are the provisions that define different property types (such as communal and public property, in addition to private property) and essentially provide a narrow definition of the private property protected by the takings clause. These provisions are typically found in a section with general principles and policies and not the bill of rights itself, and are thus separated from the main takings clause. Also suspicious are provisions that place various restrictions on landownership or that contemplate land reform. To a careful reader, these provisions reveal that land titles might be less secure than the general takings clause suggests. Policies related to land reform are usually not part of the bill of rights, but are typically placed in a section that sets forth principles for state policy. Another category of suspicious provisions comprises those that provide a separate property regime for foreigners. These provisions do not deprive foreigners of their property entirely, but subject them to a different legal regime that could make them more vulnerable to expropriation. The constitutional arrangements for foreigners, again, are not usually placed in the bill of rights but in a section on government policy.
Potentially also suspicious are limitations clauses that state that property should serve the common good. These provisions are different from the public interest requirement discussed in Part II.A in that they do not require that expropriation should serve a public interest, but rather make a sweeping statement that all property is subordinate to the public good. These clauses therefore have the potential to undermine property protections entirely: whenever a government declares something to serve the common good, private property can be curtailed. These statements are fairly common: today, they are enshrined in about one-third of all constitutions with a takings clause. What is more, these statements are usually included in the takings clause itself, making it more difficult to regard them as "fine print" that was added to the constitution to undermine the main takings clause. For these reasons, the common good clauses are not classified as "fine print," at least not for the initial exploration of the data 102 .
Another design feature for which it is unclear whether or not to treat it as fine print is the location of the main takings clause. Specifically, although most constitutions include the takings clause in the bill of rights, some place it in a different part of their constitution, such as the preamble. On the one hand, placing the takings clause outside the bill of rights could make it easier for states to circumvent the takings clause, as usually only the rights enshrined in the bill of rights are subject to judicial review. On the other hand, a number of countries place all constitutional rights in their preamble, making it difficult to predict whether this is indeed a strategy to circumvent the takings clause. For this reason, the location of takings clauses is not treated as fine print, at least not for the initial exploration of the data. Notably, the findings presented in the next Part do The Politics of Takings Clauses not depend on these classification decisions and they remain similar when reclassifying either or both of these features (that is, the common good clause and the location of the takings clause) as fine print 103 . Figure 6 depicts the prevalence of each of the main fine print provisions: (1) different types of property, (2) land restrictions, (3) land reform, and (4) separate property regimes for foreigners. It reveals that none of these provisions are very common, and that most appear in somewhere between 10% and 20% of all constitutions. Notably, many of the constitutions that include fine print include more than one of these provisions. The graph also reveals an upward trajectory for all features from the 1960s to the 1980s, with the fine print provisions reaching the height of their popularity in the late 1980s, and a decline in their prevalence from the 1990s onwards. This trajectory suggests that the fine print provisions are linked to the rise and fall of socialist constitutions. At the same time, the enduring prevalence of such provisions until this date suggests that they are not simply a feature of socialist systems but withstood the "end of history," 104 and remain prevalent in the constitutions of purportedly free market economies today. It is noteworthy that fine print provisions are often accompanied by a generous takings clause. Indeed, among the countries that include one or more of the fine print provisions identified above, the average score on the expropriation index is above the global average. Over the course of the period 1946-2013, the average score on the expropriation index was 3.21. Among countries with fine print provisions in their constitution, by contrast, it was 3.31, while among countries without fine print it was 3.16 105 . This suggests that these provisions reflect genuine contradictions in the constitution's design.

III. FROM DE JURE TO DE FACTO
A comparison of how takings clauses measure up with actual respect for property rights requires data on de facto expropriation risks in each country. A commonly used measure to capture de facto respect for private property is the "risk of expropriation index" from the International Country Risk Guide (ICRG) 106 . The ICRG is prepared by the Political Risk Services (PRS) Group, a commercial provider of political and country risk forecasts to international investors 107 . The ICRG consists of a range of variables that capture monthly information on the political, economic, and financial risks associated with investing in any given country. It is compiled by a team of country analysts that draw on a range of different sources 108 . Importantly, the ICRG assesses actual investment risks in a country and not constitutional takings clauses. This study uses the ICRG variable that specifically captures the risks of expropriation, and that does not include other types of political risks faced by investors. This variable is based on an expert assessment of how secure private property is from government intervention 109 . It ranges from zero to ten, with zero indicating the lowest level of protection against government expropriation. A cursory exploration of the data suggests that there is virtually no relationship between the de jure expropriation index and the de facto risk of expropriation. The correlation coefficient between the two variables is -0.028, which suggests that the two measures are largely unrelated, and that, if anything, their relationship is negative, meaning that a higher score on the de jure expropriation index is associated lower de facto protection against government expropriation. Figure 7  109 but this is likely to be highly correlated with property in general. Indeed prior studies use the index to measure property rights in general rather than investors' property rights investment risks on a monthly basis. I create an annual expropriation risk measure by taking the average of the monthly data. The PRS Group changed its methodology for creating the expropriation risk variable in the late 1990s. At that point, the expropriation risks variable was renamed to "contract viability," and broadened in scope to include the viability of contracts with the government in addition to risk of expropriation. The original variable covers the period of 1984-1997, while the new variable is available from 2001 to post-1997 dummy variable that controls for changes in coding methodology.
A substantial number of countries are clustered in the top left quadrant of Figure 7, meaning that they place stronger restrictions on the power of eminent domain on paper than in practice. These are the countries that might have adopted takings clauses for disingenuous reasons or where the adoption of takings clauses was surrounded by substantial societal disagreements. Another group of countries, however, is clustered in the bottom right quadrant, suggesting that these countries protect property de facto even though they place only few constitutional restrictions on the power of eminent domain. Although the theoretical discussion thus far has focused mainly on the first group (the "false positives," that is, countries that offer higher levels of protection on paper than in practice) the empirical analysis in Part IV.B will provide a preliminary exploration of the second group (the "false negatives," that is, countries that offer higher levels of protection in practice than on paper). Conversely, a lack of statistically significant correlation between constitutional takings clauses and de facto expropriation risks suggests that it is unlikely that they are related.
To explore the relationship between de jure takings clauses and de facto expropriation risks, I estimate a simple linear regression model with robust standard errors clustered at the country level, 114 country fixed-effects, and a set of predictor variables that are common in the literature.  , respectively. The graph should be interpreted as follows: we are 95% certain that the true value of the parameter lies somewhere on the depicted line. The graph reveals that there appears to be no observable relationship between de jure constitutional takings provisions and de facto expropriation risks. Failure to reject the null hypothesis-that constitutional takings clauses do not reduce de facto expropriation risksdoes not necessarily offer evidence for the null hypothesis. Just because the de jure expropriation index is not statistically significantly associated with de facto expropriation risks, this does not necessarily mean that the variable has no effect 119 . Specifically, it is possible that the effect of this variable is substantively large but fails to achieve statistical significance because the model is imprecisely estimated. This does not appear to be the case here: Figure 8 reveals that the coefficient of the expropriation index is close to zero and that the confidence intervals also fall around zero (95% confidence intervals: -0.13 to 0.22). This suggests that the effect of de jure takings clauses is not only statistically insignificant but also small 6fbWkpxTDZCQ01jYnc in substantive terms. Thus, placing more restrictions on the power of eminent domain does not appear to improve property protections in practice.
By contrast, democracy, GDP per capita, and judicial independence are all statistically significantly associated with de facto expropriation risks. The findings that democratic countries, richer countries, and countries with more judicial independence do a better job at protecting private property are all consistent with the existing literature 120 . Not only is the overall de jure expropriation index unrelated to actual property protections, but the same is true for the individual components. When correlating each of the individual components of the expropriation index with de facto expropriation risks (depicted in Table 1), most correlations are close to zero and, in some cases, negative. The one exception is for provisions that unambiguously require "full" or "market value" compensation. The correlation between this provision and the de facto expropriation index is 0.18. However, this provision is not statistically significantly associated with de facto expropriation risks after controlling for confounding factors in the same regression model reported in Figure 8 122 . In sum, the data reveals that constitutional restrictions on the power of eminent domain do not appear to affect de facto expropriation risks.

IV. CONDITIONAL EFFECTS: FALSE POSITIVES AND FALSE NEGATIVES
Although the substantive and procedural restrictions on the power of eminent domain appear to have little impact on de facto expropriation risks, it is possible that they affect government behavior only under certain conditions. This Part will first explore the theoretical explanations developed in Part I, which all concern "false positives," that is, countries that adopt takings clauses in the absence of a genuine commitment to protect private property in practice, either because the commitment is entirely false or because it is surrounded by societal disagreements. This Part will next provide a preliminary exploration of possible "false negatives," that is, countries that are committed to private property in practice but do not enshrine strong restrictions on eminent domain in their constitution.

A. False Positives
One possible explanation for the lack of correlation between de jure and de facto property protections is that some constitutions are simply disingenuous. When constitutional promises are false, the mechanisms that are supposed to make it harder to deviate from the constitution's promises ex post are likely to be absent 123 . If not, the government risks that the false promises will be turned against it 124 . One of the most important mechanisms to make constitutional promises binding is judicial independence. As noted in Part I.B, an independent judiciary can act as a neutral arbiter in property disputes between the government and private parties and can strike down unconstitutional legislation. It is no surprise, therefore, that investors have long used judicial independence as a proxy for whether their investments are secure. To explore whether the effects of takings clauses depend on judicial independence, I re-estimate the model reported in Figure 8 but with a more limited sample that only includes the 40% of countries that enjoy the highest level of judicial independence according to the CIRI human rights dataset. The CIRI data codes judicial independence on a three-point scale-full independence, moderate independence, and no independence-based on the annual U.S. State Department Country Reports on Human Rights, the input for which is collected by American embassies around the world 125  high levels of judicial independence 126 . The coefficient of the expropriation index is close to zero and not statistically significant. By contrast, democracy and GDP per capita are again positive and statistically significant predictors of de facto protection against expropriation 127 . Thus, even when omitting those states that adopt takings clauses in the absence of an independent judiciary, de jure restrictions on eminent domain and de facto expropriation risks appear to be unrelated to each other 128 . 126 Also when limiting the sample to the 75% of countries with either high levels of judicial independence or medium levels of judicial independence (omitting only the 25% of countries with the lowest levels of judicial independence), constitutional takings independence and grant the judiciary the power of judicial review (as coded by Ginsburg & Versteeg, supra note 122). Second, I limit the sample to countries that have the highest or medium levels of judicial independence and grant the judiciary the power of judicial 127 128 Although judicial independence is the main focus of the comparative constitutional law literature, it is not the only mechanism that can make constitutional constraints binding. Another potential mechanism is democracy. Constitutional commitments can be enforced through the democratic process, especially when these commitments are authoritarian regimes might adopt takings clauses for disingenuous reasons, suggesting that takings clauses in authoritarian regimes will fail to reduce expropriation risks. To explore democracy's impact, I re-estimate the model reported in Figure 8 for a subsample of democratic and autocratic countries respectively. Countries with a polity score of six or higher are considered to be democratic for this purpose. This threshold is commonly used in the literature. See, e.g., Simmons, supra note 15, at 83; Law & Versteeg, supra note 46, at 178. In neither one of these samples, however, is the de jure expropriation words, takings clauses in democratic and autocratic constitutions alike fail to reduce de facto expropriation risks. For the subsample of democratic countries, observations (N) = 1319 and the r-squared is 0.64. For the subsample of autocratic countries, N = 1204 and the r-squared is 0.59. When classifying countries with a polity2 score of 4 or higher as democratic, the results are the same. A second explanation for the lack of correlation between de jure and de facto property protections holds that, in some countries, constitutional restrictions on the power of eminent domain might enjoy only partial popular support. Where the constitution places stronger restrictions on the power of eminent domain than the level of restrictions preferred by a majority of people, political majorities might use their political clout to undermine the constitution's takings clause. It is difficult to assess whether constitutions are elite products that deviate from popular opinion; such an assessment requires in-depth case studies, which are beyond the scope of this Essay 130 . Although it is difficult to establish which groups were most influential in writing the constitution, it is relatively easy to detect the ambiguities that are built into the constitution's design. As discussed, numerous constitutions include ambiguities and contradictions that lend themselves to quantitative coding.
To explore the impact of such ambiguities, I re-estimate the model reported in Figure 8 but with a more limited sample that only includes the 77% of countries that do not include any fine print in their constitutions at all. The results, depicted in Figure 10, suggest that when accounting for the constitution's fine print, there does exist a positive and statistically significant relationship between de jure takings clauses and reduced de facto expropriation risks. In the subsample of countries without fine print, the de jure expropriation index is positive and statistically significant, albeit only at the 10%-level 131 . Specifically, a 1-point increase on the expropriation index (that ranges from 0 to 7, and for which 7 denotes the highest level of protection) is associated with a 0.18-point increase on the de facto risk of expropriation measure (that ranges from 0 to 10, and for which 10 denotes the highest level of protection). By contrast, in the subsample that only includes the constitutions with fine print (reported in Figure 11), the original expropriation index has a negative effect, although not one that is statistically significant 132 . In other words, these findings suggest that when constitutions unambiguously protect private property, more restrictions on the power of eminent domain are correlated with a statistically significant decrease in expropriation risks. Yet where constitutions include "fine print" ambiguities, which can potentially water down restrictions on the power of eminent domain, de jure takings clauses do not appear to cause a reduction in expropriation risks. 131 supra notes 144-46 and accompanying text. 132 provisions is present in the constitution, but that takes its original value otherwise. When doing so, there is a positive relationship between de jure takings clauses and reduced de facto risk of expropriation, but the p-value is 0.15, suggesting it falls just outside These effects become even stronger when adding two constitutional features for which it was not a priori clear whether they should be classified as "fine print" to the group of constitutions that include fine print: (1) limitation clauses that make property subordinate to the common good and (2) constitutions that place the takings clause outside the bill of rights. To explore the impact of these features, I first repeat the analysis reported in Figures 10 and 11 but add the constitutions that place the takings clause outside the bill of rights to the subset of constitutions with fine print. When doing so, the de jure expropriation index is positive and statistically significant at the 5% level in the subsample of countries without fine print in their constitutions, while it is negative but statistically insignificant in the sample of countries with fine print 135 . I next repeat the analysis reported in Figures 10 and 11 but add the clauses that subordinate private property to the common good to the definition of fine print. When doing so, the de jure expropriation index is positive and statistically significant at the 5% level in the sample of countries without fine print in their constitutions, while it is negative but statistically insignificant in the sample of countries with fine print 136 . Finally, I repeat the analysis reported in Figures 10 and 11 but add both types of provisions to the definition of fine print. Also in this case, the de jure expropriation index is positive and statistically significant at the 5% level in the subsample of countries without fine print in their constitutions and negative but not statistically significant in the sample with fine print 137 . This suggests that the overall finding that societal disagreements matter is robust compared to alternative definitions of what constitutes a constitution's fine print. Indeed, the various provisions that capture fine print are all positively correlated with 135 0.18 in Figure 10). 136 0.18 in Figure 10). 137 0.18 in Figure 10). each other 138 . Taken together, these findings suggest that when constitutions include contradictory property provisions, the exceptions become the rule, and the main takings clause will be undermined in practice. At the same time, when countries commit to protect private property without caveats, constitutional restrictions on eminent domain do appear to be correlated with de facto expropriation risks. Of course, the analysis does not conclusively establish that constitutional contradictions indeed result from societal disagreements. Nor does it sufficiently address possible endogeneity of takings clauses to draw any definite conclusions about the causes of these clauses' impact. Yet, this Essay's findings at least tentatively suggest that the politics at the time of constitution-writing could be a crucially important predictor of takings clauses' future performance.

B. False Negatives
The analysis thus far has focused exclusively on "false positives," that is, countries that adopt constitutional takings clauses without a genuine commitment to respect private property in practice 139 . The focus on false positives follows logically from the theoretical literature on property rights, which emphasizes the economic benefits associated with constitutional restrictions on the power of eminent domain. The literature therefore predicts that countries will be incentivized to constitutionally restrict eminent domain power, and invites a focus on why such restrictions do not work in practice. Yet it is possible that the lack of correlation between de jure and de facto property rights is in part explained by what Professor Simmons describes as "false negatives," that is, countries that respect private property in practice and yet do not place constitutional restrictions on the power of eminent domain 140 . There is no clear theoretical foundation for this phenomenon; the conventional wisdom holds that countries 138 print provisions suggests that one underlying factor accounts for all of the shared variance reveal that for this data, one factor should be retained. See DEVELLIS, supra note 107, at 128-32. 139 See SIMMONS, supra note 15, at 17-18. 140 Id. that genuinely intend to respect and protect private property benefit from adopting constitutional takings clauses. Constitutional takings clauses, after all, signal credibility to investors, capital markets, and citizens alike, which might help a government to reap the economic benefits associated with secure property rights. Yet, Figure 7 suggests that at least some countries overperform relative to their constitutional promises. One possible group of countries that might be reluctant to enshrine takings clauses into their constitutions, even when they do respect private property in practice, are the common law countries. Common law countries typically have a strong tradition of judicial interpretation of constitutional text, and may therefore omit some of the more specific restrictions on the power of eminent domain from the text itself. To explore this possibility, I estimate the same baseline model as reported in Figure 8, but exclude the common law countries from the sample. This exercise reveals no evidence that common law countries drive the results. Also in this restricted sample, the de jure restrictions on the power of eminent domain do not explain de facto expropriation risks 141 . In other words, even when excluding common law countries as possible false negatives, there is no relationship between constitutional takings clauses and reduced expropriation risks.
A second possible group of "false negatives" are those countries that possess "modest" constitutions 142 . Modest constitutions are characterized by an absence of rights protections in the constitutional text even when those rights are protected in practice. The phenomenon of the modest constitution suggests that some countries are generally reluctant to enshrine rights in their constitution, which might also affect the degree to which they adopt property rights. Law and Versteeg classify a portion of the world's constitutions as modest based on their assessment of each constitution's performance on fifteen individual rights (not including property rights) 143 .
To explore whether modest constitutions affect the results, I re-estimate the model reported in Figure 8, but exclude all the modest constitutions from the sample. This exercise reveals that even in this more restricted sample, de jure restrictions on the power of eminent domain and de facto expropriation risks are not statistically significantly correlated with each other 144 . Overall, this initial analysis reveals no evidence to suggest that the disconnect between de jure and de facto property rights is caused by false negatives.

Conclusion
There exists a longstanding debate over whether constitutions are more than mere parchment barriers. Though the debate traces back to at least James Madison, remarkably little is known about the circumstances under which constitutions actually matter. This Essay has offered an initial empirical exploration of this question in the context of constitutional takings clauses. It finds that takings clauses seen in isolation convey little information on the degree to which private property is in fact secure from government expropriation. Indeed, even takings clauses that are interpreted and enforced by an independent judiciary do not seem to reduce de facto expropriation risks. Yet, when taking into account a constitution's fine print, there does appear to be a relationship between a de jure restrictions on expropriation and de facto expropriation risks. It is important to note that this Essay does not exhaust the topic. Although it establishes with some certainty that there is no relationship between de jure and de facto expropriation risks in the aggregate, it has not explored all possible conditions under which takings clauses might make a difference. In recent years, there has been a growing consensus that when empirically exploring the effectiveness of de jure rights, researchers should not focus on aggregate effects in all countries, but rather explore the conditions under which rights do matter 145 . This Essay has merely scratched the surface of such an inquiry by examining a limited number of possible conditions relating to judicial independence and ambiguities in a constitution's design. Future research is needed to delve further into the conditions under which takings clauses might make a difference in practice 146 . Even with those limitations in mind, this Essay reveals an important theoretical point. With its focus on the politics surrounding a constitution's adoption, this Essay fits with what Professor Hirschl has dubbed the "realist" tradition in comparative constitutional law, that is, a growing number of studies that explore how politics affect a constitution's substance 147 . Well-known studies in the realist tradition have explored how elites, ruling coalitions, or both were able to shape a constitution's design to fit their own preferences. Some qualitative studies have even focused specifically on how societal disagreements produce contradictions and ambiguities 148 . Yet, few studies have connected the politics of constitution-making to the future operation of the constitution. While this short Essay leaves many questions unanswered, it suggests that the political conditions surrounding a constitution's adoption could crucially impact its future operation. It is possible that this dynamic is not unique to takings clauses alone, but also characterizes the operation of other constitutional commitments. More research is therefore needed on constitutions' fine print and, more generally, on how the circumstances surrounding a constitution's adoption affect its future operation. Indeed, it is possible that one of the most important predictors of constitutional success is not the substance of the document, but the conditions under which it was adopted.