Introduction
The law of foreign sovereign immunity changed dramatically during the twentieth century. The starting point was the rule of absolute immunity. Under that rule, a state was immune from suit in another state's courts (jurisdictional immunity) and its assets could not be seized to enforce a court judgment (execution immunity). Over time, this doctrine gave way to the so-called restrictive theory of immunity, under which foreign sovereigns were no longer immune in cases arising out of commercial (as opposed to public, or governmental) acts. The shift from absolute to restrictive immunity is one of the most significant developments in the field of customary international law (CIL) (Nagan and Root Reference Nagan and Root2013; Verdier and Voeten Reference Verdier and Voeten2015). It represented a great “tectonic shift” (Koh Reference Koh2011, 1143) in the willingness of national courts to enforce private claims against foreign governments. But despite its perceived significance, the shift is conventionally understood only as a change in the default rule (see, e.g., Verdier and Voeten Reference Verdier and Voeten2015, 3). Under absolute immunity, a creditor who wants legal enforcement rights must bargain for them (typically ex ante). Under restrictive immunity, no such bargain is necessary.
The conclusion that sovereign immunity rules are defaults emerged from a methodology commonly employed in studies of CIL. Simply defined, CIL is “general practice accepted as law” (Statute of the International Court of Justice Art. 38, April 18, 1946). In other words, a CIL rule exists if a sufficiently large number of states engage in a practice because they view it as obligatory. To identify state practice with regard to sovereign immunity, tribunals, practitioners, and scholars have examined the law on the books—that is, the formal statements of law articulated by legislatures, courts, and other government agents (see, e.g., Fox Reference Fox2008; Jennings and Watts Reference Jennings and Watts2008; ICJ Judgment of Feb. 3, 2012 [Germany v. Italy]; Verdier and Voeten Reference Verdier and Voeten2015). As a practical matter, this methodology tempts observers to treat judicial opinions, statutes, and other traditional legal materials as conclusive evidence of state practice.
We adopt a different perspective. Rather than look to the law on the books, we examine the behavior of actors directly affected by the rules of foreign sovereign immunity. In our context, which involves public debt, these actors include the sovereign governments who borrow money and the private investors who buy government bonds. The relationship between formal law and behavior is a central inquiry in sociolegal studies. Among others, Macaulay (Reference Macaulay1963), Greif (Reference Greif1989), Ellickson (Reference Ellickson1994), Bernstein (Reference Bernstein1992), George, Gulati, and McGinley (Reference George, Mitu and McGinley2011), and Hadfield and Bozovic (Reference Hadfield and Bozovic2016) have demonstrated that private actors do not always order their affairs in accordance with formal legal rules. Yet mainstream legal analysis often ignores these insights, including their potential relevance to the study of CIL.
In this article, we draw on what is (to our knowledge) the most comprehensive dataset of contracts between private citizens and foreign governments. Our data consist of sovereign bonds—that is, loan contracts sold by governments to investors in foreign markets. Because sovereign borrowers often defaulted, and because the borrowers' own courts would likely prove inhospitable or ineffective, foreign investors had an interest in accessing the courts of their home states, if the law of sovereign immunity would allow this. The dataset includes thousands of contracts and spans two centuries and multiple jurisdictions with different sovereign immunity rules, and we use it to study how loan contracts responded to changes in these rules.
We review the data below, but the principal finding is this: until the 1970s, virtually no contract, in any jurisdiction we have identified, modified the background rules of sovereign immunity. Then, in the mid- to late-1970s, everything changed. Almost overnight, bonds issued in multiple jurisdictions uniformly adopted detailed clauses bestowing legal enforcement rights on creditors. This shift correlates almost perfectly with two statutory developments: the Foreign Sovereign Immunities Act (FSIA) in the United States (1976) and the State Immunities Act, 1978 (SIA) in the United Kingdom. Every prior development in the law of sovereign immunity seemingly passed unnoticed. Prominent judicial opinions in major creditor jurisdictions; draft treaties purporting to codify CIL; official proclamations of national policy by executive branch officials—none appear to have had any impact on contracts. By contrast, the FSIA and SIA were associated with a dramatic and immediate change in contracting behavior.Footnote 1
This empirical picture is puzzling for two reasons. First, it is at odds with the conception of sovereign immunity as a default rule. Under absolute immunity, creditors had to bargain explicitly for court access, but they did not do so despite having economic incentives to seek greater legal enforcement rights. Under restrictive immunity, creditors presumptively had access to national courts, yet they bargained for access anyway. Why did express waivers of immunity begin to appear only after the shift to restrictive immunity, when they were no longer as necessary? Second, what explains the suddenness and uniformity of the change in contracting practices? Most scholars of international law describe the transition from absolute to restrictive immunity as a gradual one, taking over a century (see, e.g., Badr Reference Badr1984, 57–58; van Alebeek Reference van Alebeek2008, 151–52; Verdier and Voeten Reference Verdier and Voeten2015). This inclines observers to view statutes like the FSIA as relatively minor developments that did little more than codify existing law (see, e.g., Badr Reference Badr1984). In the context of our data, however, the gradual legal transition described by the literature on CIL seems largely irrelevant. Why, then, were the FSIA and SIA associated with such a dramatic change in market practice?
This article offers preliminary answers to these questions. In the process, it calls into question the understanding of absolute immunity as a default rule. Our findings imply that sovereign borrowers and private investors viewed immunity as akin to a mandatory rule.Footnote 2 By this, we do not mean that international law forbade national courts to give effect to a foreign government's contractual waiver of sovereign immunity. Our claim is that commentators mistakenly inferred from a relative handful of judicial decisions that courts would predictably give effect to such waivers, when these decisions were in fact tentative and ambiguous commitments that gave investors little assurance that courts would be receptive to their claims. Put another way, market participants did not interpret these early judicial pronouncements as credible commitments that courts would entertain suits against foreign governments. The FSIA and SIA, by contrast, may have represented the first credible commitments to enforce privately negotiated waivers of immunity.
Yet there is almost surely more to the story, for not all credible changes in the legal rule dramatically impact behavior. Thus, after recasting absolute immunity as akin to a mandatory rule, we turn to the impact of the FSIA and SIA on contract drafting. As we explain, actors in the sovereign debt markets are likely to value new terms only when they believe that others share that view. In consequence, a new contract clause will become widespread only if market actors believe that others will embrace the clause. We argue that the FSIA and SIA served as coordinating devices around which lawyers, bankers, and finance officials developed common expectations about how to prepare sovereign bond documentation (see, e.g., McAdams Reference McAdams2000).
We close by highlighting the implications of our findings and our methods for the CIL inquiry into state practice. As we have noted, the traditional inquiry seeks to infer state practice from judicial opinions, legislative acts, diplomatic acts and protests, and other formal acts by state agents. This approach raises serious methodological difficulties, not the least of which is that tribunals, lawyers, and legal academics rarely have the resources and disciplinary training to investigate, interpret, and aggregate such materials from nearly 200 states (Kelly Reference Kelly2000; Young Reference Young2002; Petersen Reference Petersen2009; Choi and Gulati Reference Choi, Gulati and Bradley2016; Scoville Reference Scoville2016). Our findings highlight another, perhaps deeper, flaw. Traditional methods for identifying state practices cannot easily distinguish law that matters—in the sense of impacting behavior—from law that does not. By shifting our focus to the behavior of actors affected by the rules of sovereign immunity, we highlight an alternative way to explore the nature of state practices. Our approach complements traditional CIL methods and can support (or undermine) the conclusions those methods yield.
The Evolution of Sovereign Immunity Law
This section provides an overview of the evolution of foreign sovereign immunity law. We focus on judicial decisions and other public statements (or applications) of the law by state actors. We also explore how commentators writing at the time interpreted these materials. Put differently, we tell the story that results from application of traditional methods used in the study of CIL. As will become clear when we turn to the data, a different picture emerges from the behavior of market actors. That fact complicates the story we relate in this section.
Because CIL rules are generalized from the practices of many states, they necessarily obscure differences across jurisdictions (Stephan Reference Stephan2010). To illustrate this, we give detail about the different evolutionary paths taken by the law in the three major financial centers for the issuance of sovereign bonds over the period we examine (roughly 1810–2010): France, the United Kingdom, and the United States. The law on the books differed substantially in these jurisdictions, and these differences should, in theory, have produced different practices with regard to contract drafting.
The Evolution from Absolute to Restrictive Immunity
Foreign sovereigns traditionally enjoyed absolute immunity from suit and from execution (Verdier and Voeten Reference Verdier and Voeten2015). The rule was subject to exceptions. For example, national courts could adjudicate questions of ownership over immovable property located in their jurisdictions. As most commentators understood the law, however, a foreign sovereign could not be sued without its consent, and even if it consented to be sued, its assets remained immune from seizure.
Because it is inferred from state practices in the aggregate, CIL has been called “the generalization of the practice of states” (Fisheries Case [U.K. v. Norway] 1951, 191 [J. Read]). Put differently, CIL rules can be expressed in ways that obscure variations across states (Stephan Reference Stephan2010). To say that a sovereign may consent to suit, for instance, does not indicate whether it may revoke its consent once given. By the late nineteenth century, at least some courts had indicated that consent was irrevocable. One early French case involved litigation against the Government of Tunis.Footnote 3 The relevant bonds were issued through French banks and contained a clause providing that disputes would be adjudicated in France. The Bey of Tunis had established a commission to manage his financial affairs and argued that this divested the French courts of jurisdiction. The Civil Tribunal of the Seine rejected this argument, noting that the stipulation to jurisdiction could not “be nullified by the personal will of one of the parties” (Rochaïd-Dahdah v. Gouvernement tunisien 1888). Taken at face value, this appears to be an early, unequivocal articulation of the view that absolute immunity was a default rule that could be trumped by contract.
Writing in the 1930s, Eleanor Wyllys Allen undertook a cross-jurisdictional survey of the law of foreign sovereign immunity. She discussed the Tunisian decision and other cases, noting what had become the “usual view” that a sovereign could renounce its immunity by contract (Allen Reference Allen1933, 19–20). She preferred the approach taken by English courts, which required the sovereign to consent at the time of the suit, but conceded that the law was to the contrary. Others shared her assessment. A contemporaneous proposal to codify the rules of international law on the subject (the Harvard Project) asserted that a foreign government lost its immunity from suit when it had “previously consented to the institution of such a proceeding.” The accompanying comments explained that the rule was “so obviously equitable that its general acceptability may be assumed” (Harvard Research 1932, 549).Footnote 4 After canvassing the law of multiple jurisdictions, the authors dismissed English practice as the only exception.
The rule regarding waiver of execution immunity was less well-established. Allen noted rare occasions on which a foreign state was “held to have submitted even to execution against its property” (1933, 49, 68), without suggesting that these episodes represented departures from settled law. By later in the century, there was additional authority for the proposition that a sovereign could waive execution immunity, as long as it did so expressly. A report prepared in 1949 to inform State Department deliberations on whether to adopt restrictive immunity (the Snow Report) characterized French law as receptive to such waivers, citing a case from 1938. The general belief, however, was that states probably would not waive execution immunity. Thus, while the authors of the Harvard Project implied that governments could waive execution immunity ex ante, they did not propose a rule on the subject (Harvard Research 1932, 707).
The exceptions to immunity discussed thus far all involve waivers. Increasingly, however, courts manufactured other exceptions. By the 1950s, these accrued to the point that prominent commentators expressed skepticism that absolute immunity had ever been so widely practiced as to constitute a rule of CIL (Lauterpacht Reference Lauterpacht1951). Courts had begun to distinguish acts jure imperii (i.e., in the sovereign's public or governmental capacity) from those taken jure gestionis (in a private, civil, or commercial capacity). Sovereigns retained their traditional immunity in the former category, but not the latter. This distinction became the backbone of the theory of restrictive immunity. Over the latter half of the twentieth century, states increasingly adopted the restrictive theory (Damrosch Reference Damrosch2011; Verdier and Voeten Reference Verdier and Voeten2015).
After a country adopted restrictive immunity, a foreign sovereign could be sued for its commercial acts with or without its consent, but the boundaries of commercial activity were contested. Our data, for example, consist of sovereign bonds. Early statements of the restrictive immunity rule often preserved jurisdictional immunity in cases involving public debt (see, e.g., Westlake Reference Westlake1910; Harvard Research 1932). Thus, it remained uncertain whether the switch to restrictive immunity would benefit investors in sovereign bonds. If incurring debt was a governmental rather than a commercial act, a bond investor could sue only if the bond contained (and the court enforced) a waiver of jurisdictional immunity. Moreover, even in jurisdictions that had embraced restrictive immunity with respect to questions of jurisdiction, sovereigns might retain execution immunity with respect to some or all assets. Thus, all creditors of a sovereign had reason to prefer contracts that included waivers of execution immunity.
One International “Custom,” but Varied State Practice
As noted, the conventional understanding of the evolution of sovereign immunity law obscures differences across jurisdictions. Here, we explain how the law evolved in the three most important jurisdictions for the issuance of sovereign debt: France, the United Kingdom, and the United States.
France
By relatively early in the nineteenth century, the London Stock Exchange had become the primary market for sovereign debt (Neal and Davis Reference Neal and Davis2006), with the Paris Bourse playing a secondary role. Still, Paris remained an important market into the twentieth century for Russia, China, Romania, and other borrowers (see, e.g., Lavelle Reference Lavelle2004, 35). During the period in which Paris was an important global market, French sovereign immunity law appeared to evolve. Although a civil law jurisdiction, French courts played the dominant role in this apparent evolution (Dellapenna Reference Dellapenna1992, 55).
With regard to jurisdictional immunity, we have noted that, in 1888, a French court refused to allow the Bey of Tunis to revoke consent to be sued in French courts. From this case, and a few later ones, Eleanor Wyllys Allen and other authoritative commentators confidently inferred that French law treated contractual waivers of immunity as irrevocable (Harvard Research 1932, 551–53; Allen Reference Allen1933, 168).
The law developed more slowly with regard to execution immunity. Yet if French law treated jurisdictional immunity as a default rule, it is not clear why it would adopt a different approach to execution immunity, and, in fact, early cases implied that a foreign sovereign could waive both immunities. For example, courts took pains to clarify that a generic waiver of immunity would be construed to refer only to jurisdictional immunity—an unnecessary clarification if execution immunity could not be waived at all (see, e.g., Veuve Caratier-Terrasson c. Direction générale des Chemins de fer d'Alsace-Lorraine 1885). Later commentators agreed. Describing French practice in 1949, the Snow Report unequivocally proclaimed that French courts enforced express contractual waivers of execution immunity, citing only one 1938 case. If that understanding of French law was correct, we would expect to see waivers of one or both types of immunity by the first decades of the twentieth century, if not significantly before, in bonds issued in France. (We do not.)
United Kingdom
The London Stock Exchange was the world's preeminent financial market by the early nineteenth century, and sovereign bonds began to be issued in large numbers after around 1820 (Lipson Reference Lipson1985). Most accounts trace the adoption of restrictive immunity to the enactment of the SIA in 1978. Until the SIA, moreover, it was very much in doubt whether a sovereign could irrevocably waive its immunity by contract. As noted, several English cases required the sovereign to consent at the time of the lawsuit, even if it had previously agreed to submit to the jurisdiction of English courts (see, e.g., Duff Dev. Co. v. Kelantan 1924; Kahan v. Pakistan Fed'n 1951, 1012).
The SIA changed these rules. The statute denies immunity from suit in cases where the sovereign has engaged in commercial activity and clarifies that this includes lending transactions. This makes a waiver of jurisdictional immunity of little value in debt cases, for the default rule allows creditors to sue. The SIA also authorizes creditors to seize some of a sovereign's commercial assets. However, a contractual waiver of execution immunity offers additional value by potentially expanding the pool of assets subject to execution.Footnote 5
United States
The story is more complicated in the United States, which became a major global capital market after World War I. Most accounts trace the adoption of restrictive immunity to 1952, when the Department of State (in the so-called Tate Letter) announced the executive branch's policy decision to abandon absolute immunity. However, the Tate Letter's impact was complicated, and there were other noteworthy developments both before and after 1952. For our purposes, the important point is that pre-FSIA law left unclear whether a foreign sovereign could irrevocably waive its immunity from suit or execution. Nevertheless, some prominent commentators, employing traditional methods of identifying state practice, believed that US courts would enforce immunity waivers in at least some cases.
The Tate Letter signaled the formal US embrace of restrictive immunity (Bradley and Helfer Reference Bradley and Helfer2010; Dellapenna Reference Dellapenna2011) but did not clarify the rights of investors in sovereign bonds. For one thing, the Tate Letter addressed only jurisdictional immunity. The Tate Letter also left unclear whether the issuance of bonds was a commercial activity; that question was not answered until 1992 (Weltover v. Republic of Argentina 1992). If issuing bonds was a public rather than commercial act, foreign governments would remain immune from suit notwithstanding the switch to restrictive immunity. Finally, the Tate Letter did not clarify whether a foreign sovereign could irrevocably waive whatever immunities it retained.
These uncertainties were compounded by a quirk of US practice. Although courts initially played a key role in shaping immunity doctrine, the executive branch gradually assumed primary responsibility for immunity determinations. A foreign state that wanted to claim immunity would often request a suggestion of immunity from the Department of State. If the department recognized and allowed the suggestion, its decision was effectively final, ending litigation in US courts (Weisburd Reference Weisburd1988; White Reference White1999; Bradley and Helfer Reference Bradley and Helfer2010), but if the department declined to support the claim to immunity, the court would decide the question.
This two-tiered system makes it hard to draw conclusions about pre-FSIA practice. The view that US law allowed a sovereign to revoke its waiver (see, e.g., Ku Reference Ku2014, 43) derives some support from case law (see, e.g., Beers v. State of Arkansas 1857), and from public assertions by the Department of State. In a 1961 case, Rich v. Naviera Vacuba, the department asserted that a waiver of immunity “may be revoked at will—at least prior to the time suit is actually brought” (Memorandum for the United States [In Opposition to Application for Stay of Mayan Lines, S.A.], 276, 297).
But matters were not so clear, and surveys of international and US practice before and after Rich reached the opposite conclusion. The Snow Report (Reference Snow1949), like the earlier Harvard Project (Harvard Research 1932), concluded that a waiver of immunity was enforceable under international law and did not list the United States as an exception. Early drafts of the Restatement of the Foreign Relations Law of the United States, published in 1958, also asserted that a sovereign could irrevocably waive immunity by contract (American Law Institute 1958, 194–96). Likewise, a study prepared in 1963 for the Department of State (Sweeney Reference Sweeney1963) identified the United Kingdom as the only jurisdiction where a sovereign could revoke its consent to be sued. There was, finally, reason to doubt the department would consistently intervene on behalf of sovereigns that had agreed to waive immunity. The intervention in Rich was widely viewed as motivated by political factors (see, e.g., Cardozo Reference Cardozo1963, 466–67). Indeed, less than a decade later the department reversed course, acknowledging that a sovereign could irrevocably waive immunity by contract (US Department of State Reference Sandler, Vagts and Ristau1970).
Two conclusions emerge from this murky legal landscape. First, it is easy, using traditional methods for identifying state practice, to conclude that US law was receptive to waivers of sovereign immunity by 1950 or 1960, if not before. A number of prominent commentators made precisely this inference. To be sure, the Department of State might take a contrary position in support of a foreign sovereign's claim to immunity, and in that event a court would likely defer to the department (see, e.g., Weisburd Reference Weisburd1988; Bradley and Helfer Reference Bradley and Helfer2010). However, in other cases, courts would independently decide whether to enforce a waiver of immunity (see, e.g., In the Matter of the United States of Mexico v. Schmuck 1944; Victory Transport Inc. v. Comisaria General de Abastecimientos y Transportes 1964). If this is the correct understanding of US practice, a second conclusion follows. Well before the FSIA's enactment in 1976, private creditors had reason to prefer contracts that contained an express waiver of a sovereign counterparty's immunities.
Although the FSIA clarified much about the US law of sovereign immunity, it left questions unanswered. The statute codified restrictive immunity doctrine, but it did not expressly address whether sovereign borrowing constituted commercial activity (Weidemaier Reference Weidemaier2014). To that extent, the FSIA did little more than restate pre-Tate Letter practice. The statute, however, removed the State Department from the picture by transferring immunity decisions to the courts. The statute also clarified that a foreign sovereign could not revoke its waiver of jurisdictional or execution immunity. As with the SIA in the United Kingdom, the FSIA granted creditors limited execution rights and allowed creditors to contract for greater rights by negotiating a waiver of execution immunity.
Preliminary Hypotheses
In sum, the law of foreign sovereign immunity followed different evolutionary paths in these three countries. Figure 1 summarizes these paths, focusing on the eras of primary relevance.Footnote 6 The figure does not capture all relevant developments, in particular those affecting immunity from execution. Nevertheless, it captures the most pertinent differences.
Evolution in the Law of Sovereign Immunity
If creditors were attuned to these developments and wanted to add legal sanctions to the enforcement mix, we would expect different contracting practices to have evolved in each jurisdiction.Footnote 7 In particular, we can make the following predictions:
-
1. In France, waivers of jurisdictional immunity should have begun to appear by the late nineteenth century, and waivers of execution immunity not long thereafter. Bonds issued after the transition to restrictive immunity would continue to include waivers, both to remove uncertainty about whether bond issuance was a commercial act and to make a wider range of assets subject to execution.
-
2. In the United Kingdom, waivers of jurisdictional immunity arguably should not appear at all. Pre-SIA law treated these as revocable; the SIA, by expressly stating that government borrowing was a commercial activity, made them unnecessary. Post-SIA, however, a creditor probably would have valued a waiver of execution immunity, which could no longer be revoked and expanded the pool of assets subject to seizure.
-
3. In the United States, at any point in the twentieth century, waivers of jurisdictional and execution immunity might have been enforced in at least some cases, although the likelihood of enforcement increased over time. The FSIA might have diminished their importance slightly, but not by much, since both types of waiver continued to have a purpose. By the early 1990s, however, waivers of jurisdictional immunity would no longer serve much function, as it had become clear that sovereigns lacked jurisdictional immunity in cases arising out of bond debt.
These predictions assume that formal changes in the law influence social and economic behavior. On that assumption, our central prediction is that contracting practices should vary both in time and across the key jurisdictions.
How Contracts Responded
Contract law scholarship generally pays little attention to how and why contracts change (Richman Reference Richman2011; Choi, Gulati, and Posner Reference Choi, Gulati and Posner2013). The assumption is that each contract is negotiated from scratch. Yet this assumption is incompatible with the experience of practicing lawyers, who typically start with a template and make changes only as necessary to accommodate the present transaction (see, e.g., Weidemaier, Scott, and Gulati Reference Weidemaier, Scott and Gulati2013). This reality—in which contracts are “sticky” but do change—brings different questions to the fore. One is how legal change impacts the contract template (see, e.g., Hoffman Reference Hoffman2014). In our context, the question is especially interesting, for the changes affect the law's relevance more than its content. When legal actors in a country declare that they will treat sovereign immunity as a default rule rather than a mandatory one, they invite private citizens and foreign governments to bring their disputes out of the domain of politics and into the domain of law. Our data offer a window into how contracts responded to this invitation.
The Data
A typical sovereign bond issue includes multiple documents, including some or all of the following: (1) a contract between the government and its financial intermediaries; (2) a trust indenture, if the bond calls for a trustee to represent bondholders; (3) the bond itself, governing the relationship between the government and its bondholders; and (4) a prospectus or other sales document distributed to investors before the bond issue. Sales documents typically describe or reprint the terms that will appear in the bond.
Conceptually, the dataset has two parts. The first part includes bonds issued between roughly 1820 and 1980. Modern electronic databases do not include most of these bonds; nor is there a comprehensive paper archive. Many government borrowers, however, chose to list bonds on a stock exchange, and exchange archives often include the bonds or related sales documents. We visited the archives for the New York Stock Exchange (at the Library of Congress) and the London Stock Exchange (at the Guildhall) and took digital images of sales documents.
This method of collection left gaps. For instance, some bonds were listed only on other exchanges or not listed on any exchange. Information about these bonds might exist in bank archives, but many of these have been destroyed or are inaccessible to researchers. Nevertheless, some prominent bank archives are open to researchers. We visited the Rothschild Archives, the JP Morgan archives at the JP Morgan Library and Museum, and the archives for Barings, UBS, and HSBC banks. We also visited other collections of sovereign bonds and sales documents, as well as some collections of personal papers from prominent bankers, at the Harvard Business School library, the British Museum, the Origins of Value Museum at Yale University, Cornell University, Duke University, Columbia University, and the Museum of Historical Shares and Bonds (Museum Wertpapierwelt).
The second and larger part of the dataset consists of documents gathered from public databases, which provide fairly comprehensive data from around the mid-1980s. We used Thomson One Banker and Perfect Information. Our data include 1906 bonds issued between 1822 and the present.Footnote 8 Because London and New York were the primary markets for sovereign bonds during this period, most of our bonds come from these markets. However, the dataset also includes bonds issued in the German, French, Dutch, Japanese, and other markets.
The dataset spans the two primary eras of sovereign bond lending. The first lasted from the early- to mid-1800s through the Great Depression. The second began around 1990 and continues today. Between these eras, sovereign bond markets were relatively dormant (Cassis Reference Cassis2006). Governments borrowed primarily through direct loans from multilateral institutions such as the World Bank or from commercial bank syndicates (Fisch and Gentile Reference Fisch and Gentile2004). As Figure 2 shows, the distribution of bonds in our dataset reflects this history. Figure 2 also reveals that while we have located a substantial number of bonds from the first era of sovereign bond lending, most were issued in the early 1900s. Sovereign bonds from the 1800s are somewhat underrepresented.
Total Bonds in Dataset, by Year
We believe this is the most comprehensive dataset of bond terms, but there remain gaps in the data. The most notable relates to the Paris Bourse, which, despite its relative insignificance by the early twentieth century, remained a key market for a number of sovereign issuers (Lavelle Reference Lavelle2004, 35). We gathered data on the Paris market from three sources. First, the British Museum has a large collection of bonds, many issued in Paris. Second, many bonds were listed on multiple exchanges (e.g., London and Paris). Our visits to the NYSE and the LSE captured many such issues, and the documentation usually included the terms applicable in each market. We also purchased or copied bonds issued in Paris from specialty shops in the bourse area. Through these methods, we were able to gather 126 bonds issued by foreign sovereigns in the Paris market.Footnote 9 We cannot be sure that the sample is representative of the broader Paris market but, given the consistency of our data, we do not view this as a serious limitation.
The Late-1970s Shift in the Contract Template
Coding encompassed a number of clauses that address legal enforcement. These include clauses submitting to foreign court jurisdiction or otherwise waiving jurisdictional immunity. As an example, bonds issued by Finland in 1977 provided:
Finland will irrevocably waive any immunity from jurisdiction to which it might otherwise be entitled in any action arising out of or based on the Bonds which may be instituted by any holder of a Bond in any State or Federal court in New York City or in any competent court in Finland. (Prospectus for the Republic of Finland, $50,000,000 8.75% Bonds Due 1992 [Oct. 19, 1977])
We also coded for express waivers of execution immunity,Footnote 10 and for a variety of other clauses that remove technical hurdles, such as service of process, that impede creditor litigation against foreign sovereigns.
We find that, virtually without exception, bonds did not waive sovereign immunity or address any other aspect of legal enforcement until the FSIA and SIA had been introduced and enactment was imminent. After enactment, the market turned on a dime. Whereas virtually no bond, in any market, included a waiver of sovereign immunity before the passage of these two domestic statutes, almost every bond issued post-FSIA (in the New York market) or post-SIA (in the English market) waived the issuing government's jurisdictional immunity. Figure 3 depicts this shift across all markets.Footnote 11
Jurisdictional Immunity Waivers, All Markets (1850–Present)
With one exception—five series of bonds issued simultaneously by Malaysia in 1965—there are no waivers of jurisdictional immunity in our data until 1975. In that year, Norway issued bonds in the New York market with clauses waiving jurisdictional immunity and submitting to the jurisdiction of New York courts. By this time, the FSIA was well on its way to enactment.Footnote 12 Waivers of jurisdictional immunity quickly became commonplace in the New York market. A nearly identical transition occurred in the English market beginning in 1977.
An equally dramatic, though less uniform, shift occurred with regard to waivers of execution immunity. In our data, the first government to make such a waiver was Panama, in a 1977 issuance of bonds governed by English law. Thereafter, virtually all bonds issued in the English market included a waiver of the foreign sovereign's execution immunity (94 percent). The transition occurred a bit more slowly in the New York market, where waivers of execution immunity remained rare through the 1980s. But after 1995, waivers of execution immunity became relatively commonplace (72.8 percent) in bonds governed by New York law. Figure 4 depicts the shift across the entire dataset.
Execution Immunity Waivers, All Markets (1850–Present)
These patterns are hard to square with the conventional understanding of how sovereign immunity law evolved. Again, that understanding treats the switch from absolute to restrictive immunity as a change in the default rule (see, e.g., Verdier and Voeten Reference Verdier and Voeten2015). But if that were so, we would expect waivers of jurisdictional immunity to be common during the era of absolute immunity and less common thereafter (as states clarified that the issuance of bonds was a commercial act for which foreign governments were not immune under the new default rule). Our data demonstrate the opposite pattern. As for the prevalence of clauses waiving execution immunity, we would not necessarily expect the switch to restrictive immunity to have an impact. That is because, even under restrictive immunity, such waivers can expand the range of property subject to execution. But neither would we expect waivers to appear only after the switch to restrictive immunity. And there is another puzzle: Why were waivers of sovereign immunity so quickly and universally adopted?
Before turning to these questions, we offer four caveats. First, we acknowledge that there are a few reported cases from the 1800s and early 1900s that involve bond debt, some arguably involving waivers of sovereign immunity. Indeed, these cases contributed to the views expressed by the authors of the Harvard Project (Harvard Research 1932, 551–52) and similar efforts to identify international custom regarding sovereign immunity. To take one example, in a 1922 case, a French court asserted jurisdiction over the Department of Antioqua (Columbia) in a case involving bond debt. The report of that case recounts how the government had appointed an individual to represent it in any “judicial controversy or question” arising out of the loan (Crédit Foncier d'Algérie et de Tunisie c. Département d'Antioquia 1922). Our dataset does not include the bonds at issue in this case, although no similar language appears in any of the bonds we do have.
These exceptions, however, do not materially alter the picture that emerges from our data. The number of reported “waiver” cases involving sovereign bond debt is tiny and must be contrasted with the uniformity of our data. For example, our dataset includes 111 bonds issued before 1940 in the French market. Not one includes a clause appointing someone to represent the issuing government in any litigation relating to the debt. Nor do any include a clause referencing sovereign immunity or agreeing to submit to the jurisdiction of foreign courts. Indeed, we have found no reported case involving such an express waiver of immunity. At most, these reported cases reveal that some bonds issued in the first era of bond lending included language that a court might construe to waive sovereign immunity. These reported cases do not indicate that such clauses were common; indeed, our data suggest that they were very unusual.
A second caveat relates to bilateral or multilateral treaties that include a waiver of sovereign immunity. As a historical matter, there are numerous examples of such treaties. In the 1930s, for example, the Soviet Union entered bilateral treaties with France, Germany, and other countries that allowed investors from these countries to bring suit against the Soviet Union in the courts of the investor's home state (although such treaties rarely encompassed claims arising out of bond debt). Likewise, throughout the nineteenth and much of the twentieth centuries, countries used the so-called treaty of friendship, commerce, and navigation to procure certain rights for citizens living in other countries (Coyle Reference Coyle2013). These bilateral treaties often addressed the subject of legal enforcement, although typically by giving citizens access to courts in the host state (or, later, arbitration).
This history supports the view that as a matter of CIL, sovereign immunity could indeed be waived by treaty, but this does not mean that national courts regularly enforced waivers of immunity in contracts negotiated by private creditors. That practice—so readily embraced by the Harvard Project and similar surveys—is hard to square with our data. And one can see why state practice might distinguish between the two kinds of waiver. Litigation that results from a treaty-based waiver of immunity is not likely to cause serious diplomatic concerns for the court's home state. After all, political actors in that state agreed to the treaty. To recognize contractual waivers of immunity, by contrast, is to let private creditors haul foreign governments into national courts, whatever impact this might have on diplomatic relations between the court's home state and the foreign sovereign.
Third, anecdotal evidence suggests that sovereign immunity waivers began to appear somewhat earlier in direct loan contracts involving commercial banks. Writing in 1967, Georges Delaume provided examples, one from as early as 1945 (Delaume Reference Delaume1967, 170–72), and asserted without further support that direct loans routinely included such clauses.Footnote 13 If he was correct, sovereign immunity waivers may have become commonplace in direct loans a bit earlier than in sovereign bonds.Footnote 14 By and large, however, the practices described by Delaume are consistent with our data. In particular, even direct loans did not embrace waivers of sovereign immunity until well after such waivers had ostensibly become enforceable in key jurisdictions.
The fourth caveat relates to arbitration. For more than a century, international courts and tribunals have occasionally presided over claims arising from sovereign debt obligations (Waibel Reference Waibel2013). In many of these cases, the tribunal acquired jurisdiction through treaty or executive agreement, rather than because the loan contract provided for arbitration (see, e.g., Wuerth Reference Wuerth2003, 26–27 and n176). But a handful of bonds in our dataset do include clauses providing for arbitration of debt-related disputes. Focusing on the period before 1945, only four of 534 bonds include such an arbitration clause. These bonds, however, share certain features with treaty-based waivers of immunity. In particular, the relevant loans involved the participation of other sovereign governments as potential creditors—for example, as guarantors of the loan.Footnote 15
Exploring the Impact of Sovereign Immunity Statutes
With these caveats, our data paint a picture at odds with the stylized hypotheses we set out earlier. Although sovereign immunity law developed differently across jurisdictions, market practice did not reflect these differences. Indeed, contracts were not revised in response to any of the early legal developments that led Eleanor Wyllys Allen and other commentators to conclude that absolute immunity was merely a default rule. In the sections to follow, we explore the implications of these findings.
Absolute Immunity as a Mandatory Rule
We begin by reconsidering the conventional wisdom that absolute immunity was a default rule. Our data are more consistent with the view that immunity could not be waived, or at least that investors had significant doubt as to the enforceability of contractual waivers of immunity. If that is correct, then it would be more accurate to characterize absolute immunity as a mandatory rule—or, at least, an extraordinarily sticky default. To be clear, we do not mean to imply the existence of a CIL rule forbidding national courts to give effect to a foreign government's waiver of sovereign immunity. We argue only that market practice undermines the descriptive claim underlying the view that absolute immunity was recognized and understood to be a default rule—that is, the claim that national courts would routinely enforce contractual waivers of immunity.
To begin making this case, it may help to consider why an investor contemplating the purchase of foreign government bonds might care about legal enforcement. With few exceptions (see, e.g., Bulow and Rogoff Reference Bulow and Rogoff1989; Schumacher, Trebesch, and Enderlein Reference Schumacher, Christoph and Hendrik2014), the economics and political science literature assumes that even in the modern era, legal enforcement plays little role in the sovereign debt markets (see, e.g., Panizza, Sturzenegger, and Zettelmeyer Reference Panizza, Sturzenegger and Zettelmeyer2009). The logic behind this view is that court judgments only matter when backed by threat of asset seizure, but foreign governments can easily keep assets safe within their borders.
This skepticism is justified to a degree, but it also understates the modern and historic importance of legal enforcement (Weidemaier and Gulati Reference Weidemaier and Gulati2015). Market participants and policy makers have long treated legal enforcement as a relevant concern. For instance, even during the era of absolute immunity, creditors who had not bargained for a waiver of immunity sometimes tried to assert their claims in the courts (Allen Reference Allen1933, 166). Recall, moreover, that when negotiating treaties, countries often obtained a waiver of immunity to protect their citizens' foreign investments. It is hard to see why they would bother doing so if domestic constituents did not value legal enforcement. It is also hard to explain early efforts to reform the sovereign debt markets, which sometimes emphasized the potentially disruptive effect of litigation. For example, a 1930s initiative sponsored by the League of Nations proposed reforming bond contracts to prevent “too small a number of bondholders from attempting to institute legal proceedings when the common interests of the bondholders are not really involved” (League of Nations 1939, 33). Such initiatives suggest that legal enforcement is (and was) at least peripherally relevant.
If a creditor wants to sue a foreign government, it will generally prefer to do so in the courts of its home state. This means, at minimum, that those courts must deny the sovereign's claim to jurisdictional immunity and enter a judgment for the amount due. Having obtained such a judgment, the creditor might find its position materially improved even if the sovereign retained execution immunity. The government might pay the judgment voluntarily, or political officials in the creditor's home state might view a court judgment as a basis for diplomatic intervention (Lauterpacht Reference Lauterpacht1951, 222).Footnote 16 If the sovereign also loses its execution immunity, the creditor's position will be improved further. To be sure, the creditor is not likely to seize any assets, for the sovereign can shelter them within its borders. To do this, however, the sovereign must forego putting its assets to more productive use. For example, it cannot buy commercial goods or borrow money in a jurisdiction where it lacks immunity from execution (Shleifer Reference Shleifer2003; Cruces and Trebesch Reference Cruces and Trebesch2013; Schumacher, Trebesch, and Enderlein 2014). The cost of these foregone opportunities might induce the sovereign to pay (Weidemaier and Gelpern Reference Weidemaier and Gelpern2014).
For these reasons, investors had an economic incentive to favor loan contracts that enhanced access to legal enforcement. In turn, sovereign borrowers should have been willing to confer such rights in the hope of lowering borrowing costs. This is a familiar pattern in which borrowers cede aspects of their sovereignty to provide assurances of repayment. In the nineteenth and early twentieth centuries, for example, many borrowers ceded control over ports and other revenue-collection points to agents appointed by lenders or their home states (Mitchener and Weidenmier Reference Mitchener and Weidenmier2005a,Reference Mitchener and Weidenmier2005b; Pérez and Weissman Reference Pérez and Weissman2007; Ahmed, Alfaro, and Maurer Reference Ahmed, Alfaro and Maurer2010; Maurer Reference Maurer2013).Footnote 17 If investors valued enforcement rights, there is no reason why governments accustomed to onerous contract terms would draw the line at clauses waiving sovereign immunity—if, that is, courts would predictably give such clauses effect.
But our data are more consistent with a world in which courts were reluctant to enforce privately negotiated immunity waivers until relatively late in the twentieth century. We do not claim that investors in 1910 cared as much about legal enforcement as investors in 2010. In earlier eras of bond lending, other enforcement strategies, such as coordinating to deny future loans, may have been relatively more effective. On occasion, investors also may have hoped that rich, capital-exporting countries would intervene to protect their interests. For instance, early-twentieth-century investors might have expected the US government to protect investments in Latin America (see, e.g., Rosenberg Reference Rosenberg1999; Maurer Reference Maurer2013). But such explanations do not fully account for the pattern in our data. For instance, if contracts omitted sovereign immunity waivers because investors expected rich countries to help them, then contracts should have incorporated waivers once it became clear that help was no longer forthcoming. This became clear well before the 1970s.Footnote 18 The basic puzzle, then, remains. For over a century, sovereigns and their creditors have had gradually increasing economic incentives to negotiate clauses waiving sovereign immunity. The near-universal failure to do so implies that they doubted the enforceability of such clauses.
Seeking further insight, we reviewed many of the early legal developments that led commentators to treat sovereign immunity as a default rule. Even a cursory review prompted skepticism that investors would have expected courts routinely to enforce contractual waivers of immunity. Unlike the FSIA and SIA, early inroads into sovereign immunity gave legal actors significant discretion in how to deploy the rules. As a practical matter, this often worked in the favor of preserving a foreign sovereign's claim to immunity. We have already noted one example. During the period when the Department of State was primarily responsible for making immunity determinations in the United States (roughly 1935–1975), observers believed it to be influenced primarily by foreign policy considerations rather than the law of sovereign immunity (see, e.g., Leigh Reference Leigh1969).Footnote 19
Other jurisdictions offer similar examples. With regard to jurisdictional immunity, for instance, the rule on the European continent was ostensibly clear and unqualified: a foreign sovereign could be sued when it had waived its immunity by contract (Harvard Research 1932; Cohn Reference Cohn1958, 264–65; Delaume Reference Delaume1967, 159–60). Looking more closely at the cases underpinning this rule, however, reveals that courts sometimes hesitated to intervene in lawsuits against foreign governments, often on debatable legal grounds. For example, one 1920 Czech case dismissed a lawsuit notwithstanding the submission-to-jurisdiction clause in the creditor's contract with the Austrian State Railways. In the court's view, the disruption associated with the dissolution of the Austro-Hungarian monarchy, which coincided with contract formation, meant that railway officials might not have noticed the clause. Despite this outcome, neither the Harvard Project (Harvard Research 1932) nor the Snow Report (1949) let the Czech case undermine the conclusion that absolute immunity was a default rule, although a creditor might have taken a rather different lesson from the case.
With regard to execution immunity, creditors may have been even more dubious that courts and other state actors would vigorously enforce their rights. Courts often deployed presumptions—such as the presumption that a sovereign's generic waiver of immunity extended only to jurisdictional immunity—that prevented creditors from seizing sovereign assets. (The presumption remains the law today.) To seize sovereign assets, a creditor also had to enlist the participation of other government officials. In some cases, seizure required the express approval of local officials, which was not always forthcoming (Delaume Reference Delaume1967, 205).
This background may help explain the pattern observed in our data. Take the question of why bonds issued to French investors did not include waivers of jurisdictional immunity. Although the law on the books authorized such waivers, creditors may have been skeptical that French courts would reliably assume jurisdiction over lawsuits against foreign governments. Moreover, unless the sovereign's failure to pay a judgment would have reputational or diplomatic repercussions, the creditor would also need to get around the sovereign's execution immunity. Here, French law was less clear, but French courts had demonstrated little interest in facilitating the seizure of foreign government assets (Allen Reference Allen1933, 181–85). Given this reluctance, and the fact that creditors could not have credibly threatened to seize sovereign assets without the assistance of other government officials, it is perhaps unsurprising that contracting practices in France match those in the United Kingdom, where the law on the books treated waivers as revocable.
The same may be true of contracts in the US market. By 1952, the United States had embraced restrictive immunity with regard to jurisdictional immunity. Even if debt issuance constituted a governmental rather than commercial activity, there was increasing support for the proposition that a sovereign could irrevocably waive both jurisdictional and execution immunity. Yet until 1976, the Department of State was primarily responsible for implementing these rules, and common wisdom was that it based decisions on political rather than legal considerations (Whytock and Chilton Reference Whytock and Chilton2015). Especially between 1961 and 1970—the window between the department's assertion that a sovereign could revoke its waiver of immunity (in Rich) and its subsequent embrace of a contrary rule—creditors may have doubted the value of legal enforcement. True, prominent legal authorities asserted that US law would not allow revocation of an immunity waiver, but these were even less authoritative statements of the law than in France, where early cases had articulated the rule in the context of an actual dispute.
To be clear, we are not saying that the SIA and FSIA radically improved the position of creditors. In many respects, the FSIA's practical impact was modest (see, e.g., Weidemaier Reference Weidemaier2014). If the foregoing analysis is correct, however, the legal landscape prior to the statutes was pervaded with discretion, which legal actors might exploit to avoid becoming embroiled in disputes between foreign governments and private creditors. Indeed, this tendency exists even today. In the past few years, creditors have tried but failed to seize Argentine assets, including diplomatic property, despite a contractual waiver of immunity that seems to permit seizure. Courts have avoided allowing seizure of such sensitive assets by requiring the waiver to refer expressly to the specified assets (Société NML Capital Ltd. v. République Argentine 2011; Société NML Capital [Iles Caïmans] v. Etat d'Argentine 2013).
Despite such cases, modern law leaves courts significantly less discretion in deciding whether to hear a lawsuit against a foreign sovereign or to enforce the resulting judgment. The SIA withholds jurisdictional immunity in disputes over “the provision of finance” (SIA § 3(3)), and the FSIA commands that a waiver of sovereign immunity is to be enforced “notwithstanding any withdrawal of the waiver which the foreign state may purport to effect” (§ 1605). Both statutes likewise require the enforcement of waivers of execution immunity (SIA § 13(3); FSIA § 1610(a)(1)). Developments prior to the statutes, in contrast, left courts with ample discretion, which they sometimes deployed to deny creditors access to legal enforcement rights.Footnote 20
The FSIA and SIA as Coordinating Devices
Even if we are correct thus far, the fact that post-FSIA bonds so overwhelmingly incorporated waivers of sovereign immunity calls for further explanation. Not every sovereign bond is identical, but there is pressure toward the use of standardized contracts (see, e.g., Choi and Gulati Reference Choi and Gulati2004) to facilitate pricing and investment decisions (see, e.g., Broad v. Rockwell Int'l Corp. 1981, 929, 942–43). In the sovereign debt context, standardization also may result from the fact that a small group of global law firms dominates the market for legal services (see, e.g., Bradley, Salvatierra, and Gulati Reference Bradley, de Lira Salvatierra and Gulati2014). Market concentration, combined with the tendency for lawyers to rely on existing forms (Gulati and Scott Reference Gulati and Scott2013), means that sovereign bonds are based on a handful of templates.
Despite the pressure toward standardization, it is odd that nearly every government would allow itself to be hauled into foreign courts. As noted, loan contracts have long required governments to cede aspects of their sovereignty. A common example from the era of absolute immunity involves ceding control over ports and other revenue-collection points (see, e.g., Rosenberg Reference Rosenberg1999; Maurer Reference Maurer2013). But such arrangements were generally reserved for borrowers without strong reputations for repayment. Submitting to foreign court jurisdiction may be less offensive to sovereignty than ceding physical control of sovereign territory, but it is not a matter of indifference. Creditworthy governments surely could have maintained access to capital markets without agreeing to waive sovereign immunity. So why did the FSIA and SIA prompt such a uniform shift in the contract template?Footnote 21
On this question, a growing literature on how law can facilitate coordination may offer insight (see, e.g., McAdams Reference McAdams2015). The actors involved in sovereign debt transactions have a keen interest in coordinating their behavior. As one example, government finance officials may worry that investors will interpret the use of a nonstandard clause to signal negative information about the government's creditworthiness. Thus, finance officials long resisted the use of clauses designed to facilitate debt restructuring, often expressing the concern that investors would expect a premium to hold securities with the new clauses (see, e.g., Gelpern and Gulati Reference Gelpern and Gulati2006). But such concerns are eased if governments can coordinate their moves to new contract templates. In a market-wide, coordinated switch, no government sends an adverse signal by following the herd.Footnote 22
When parties have an incentive to coordinate their behavior, the law can play a facilitative role. One way this happens is that legal rules can influence expectations about how others will behave. As an example, consider the rule “drive on the right” (see, e.g., Schelling Reference Schelling1960; McAdams Reference McAdams2000). A third party who expresses this rule can facilitate coordination even if the party has no sanctioning power or claim to legitimacy (McAdams Reference McAdams2000). Simple expression of the rule can enable coordination by creating the expectation that others will follow the rule.
This is not to say that the identity of the party expressing the rule is irrelevant. When the party has authority as a rule maker, its pronouncements should more effectively guide behavior. The same is true when a party is empowered to impose sanctions for violation of the rule. For these reasons, courts and legislatures may be especially good at facilitating coordination (see, e.g., Garrett and Weingast Reference Garrett, Weingast, Goldstein and Keohane1993; McAdams Reference McAdams2005). Some actors are also effective aggregators of information about behavior. When such actors assert a rule, this may prompt observers to update their beliefs about the prevalence of desirability of a behavior. For instance, a legislative prohibition on smoking in public may cause observers to update beliefs as to the dangers of second-hand smoke (Dharmapala and McAdams Reference Dharmapala and McAdams2003).
In our context, of course, the pre-FSIA market was already coordinated around a contract template that did not waive sovereign immunity. So the question is why the FSIA and SIA prompted a coordinated shift to a new template. If we are correct that the statutes represented the first credible commitments to enforce sovereign immunity waivers, then part of the answer is surely that lawyers had previously seen little point in drafting such clauses. It is also important to recognize that the statutes fundamentally disrupted the contracting status quo. Recall that the FSIA and SIA gave creditors new enforcement rights by default, including limited rights to execute a judgment against sovereign assets. Because of this, a government borrower that wanted to maintain the functional status quo—in which it had near-total immunity—could not keep using the same contract template. It would have to contract around the new rules by, for example, demanding that investors bring suit exclusively in its domestic courts.
By codifying restrictive immunity, then, the FSIA and SIA effectively forced government borrowers and investors to reflect on what role legal enforcement should play in the sovereign debt markets. We speculate that lawyers may have played an important role in shaping the answer. For them, the FSIA and SIA were significant developments that made questions of legal enforcement especially salient (Weidemaier Reference Weidemaier2014). The statutes were prominent topics of discussion and provoked debate about the rules appropriate for sovereign debt cases (see, e.g., Delaume Reference Delaume1973). It would have been impossible for this relatively small group of lawyers not to consider the impact of new sovereign immunity rules on bond documentation. And because sovereign immunity waivers were now unequivocally enforceable, it would have been difficult to justify omitting such a clause given the (relatively) creditor-friendly new default rule. Finally, given the concentrated legal market, lawyers would have been good aggregators of information, credibly able to represent that sovereign immunity waivers would become the new market standard. In this way, lawyers could assure government finance officials that they would not be adopting an idiosyncratic clause.
To conclude, a number of factors likely explain why the FSIA and SIA had such a dramatic impact on bond contracts. As credible commitments to enforce immunity waivers, the statutes arguably represented a significant change from prior law. By changing the default rule to provide creditors with limited enforcement rights, the statutes made it difficult for government borrowers to bargain for a return to the status quo. By increasing the salience of legal enforcement, the statutes made it difficult for lawyers to ignore sovereign immunity when documenting a bond issue. And because of the concentrated nature of this legal market, lawyers were well-positioned to overcome the natural reluctance among finance officials to expose their governments to litigation in foreign courts. If this explanation is correct, it should not be surprising that the FSIA and SIA had a dramatic impact on market practice.
CIL and the Search for State Practice
In this final section, we return to our starting point: the topic of CIL, defined as “general practice accepted as law” (ICJ Statute, Art. 38). The definition incorporates two elements. The first is state practice. When there is a general and consistent practice of states, these “international behavioral regularities” (Goldsmith and Posner Reference Goldsmith and Posner1999, 1116) can crystallize into a binding legal rule. This occurs, however, only if states view the practice as obligatory (see, e.g., Guzman Reference Guzman2005). This second element—the belief that the practice is legally required—is called opinio juris.
We are particularly interested in the state practice requirement. International law scholarship already recognizes the difficulties associated with identifying when states engage in a general and consistent practice. For instance, CIL rules ostensibly emerge only when state practice is “uniform, extensive, and representative” (International Law Association 2000, 20). Yet this does not mean the practice must be universal (see, e.g., Trimble Reference Trimble1986; Crawford Reference Crawford2008). Indeed, CIL rules can arise even if affected states sometimes act inconsistently with the purported rule. This begs the question of how much inconsistency will be tolerated, and here opinions diverge (see, e.g., Guzman Reference Guzman2005). Similar uncertainty plagues the question of how much time must pass before state practice crystallizes into CIL (see, e.g., D'Amato Reference D'Amato1971; Roberts Reference Roberts2001).
These and other doctrinal uncertainties have prompted criticism of CIL (see, e.g., D'Amato Reference D'Amato1971; Trimble Reference Trimble1986; Goldsmith and Posner Reference Goldsmith and Posner1999; Kelly Reference Kelly2000). Despite its critics, CIL remains a vital source of international law, one that binds even dissenting states in most cases.Footnote 23 And the state practice element remains an important aspect of CIL analysis. To determine whether a particular rule has attained the status of CIL, students and practitioners of international law must make some effort to determine whether states behave in ways that support or undermine the rule.Footnote 24
The first step in such an inquiry is to identify relevant practices. Most lists would include national legislation, decisions of international and national tribunals, executive orders, and treaty commitments (see, e.g., Kelly Reference Kelly2000; Verdier and Voeten Reference Verdier and Voeten2015). Reaching beyond these core examples, a more inclusive list might include:
diplomatic correspondence, policy statements, press releases, the opinions of government legal advisers, official manuals on legal questions (e.g. manuals of military law), executive decisions and practices, orders to military forces (e.g. rules of engagement), comments by governments on ILC drafts and accompanying commentary …, an extensive pattern of treaties in the same terms, the practice of international organs, and resolutions relating to legal questions in UN organs, notably the General Assembly. (Crawford Reference Crawford2008, 24)
It should go without saying that this inquiry raises potent methodological difficulties. As Guzman (Reference Guzman2005, 126) puts it:
[I]t is fantastical to think that lawyers in a case, much less adjudicators deciding a case or policymakers selecting a course of action, can canvass the virtually infinite universe of potential evidence, let alone come to some understanding of the extent to which a practice has been followed.
Such criticisms imply a danger inherent in efforts to identify general and consistent state practices: Without the ability (and perhaps the inclination) to carefully examine how state actors behave, there is a risk that tribunals and scholars will let normative preferences trump actual state assent (see, e.g., Weisburd Reference Weisburd2015).
Our project lends some credence to this skeptical view. Indeed, one implication of our findings is that matters may be worse than they seem. Even the most extensive and rigorous surveys of state practice (see, e.g., Verdier and Voeten Reference Verdier and Voeten2015) tend to focus on judicial opinions and other formal statements or applications of the law (e.g., statutes, regulations). Of course, such materials do constitute evidence of state practice. Our point, which we think uncontroversial, is only that formal statements of the law are often much more equivocal commitments than is apparent from their text. Yet traditional CIL methods tempt researchers to overlook this fact and to assign formal legal materials undue weight as evidence of how states actually behave.
This article has focused on a particular example of state practice: that of giving effect to a waiver of sovereign immunity in a private creditor's contract with a foreign government. By reading cases and other traditional legal materials, prominent authorities writing in the early and mid-1900s concluded that the general practice was to enforce such waivers, with the United Kingdom as perhaps the only clear exception. They apparently did not think to ask whether these judicial developments represented credible commitments to respect immunity waivers in the future. This is not because the question is unimportant as a matter of CIL doctrine. To the contrary, it is clear that “cheap talk” merits little weight in the state practice inquiry (see, e.g., International Law Association 2000, 13–14). An equivocal statement of a legal rule—made so as not to constrain future state officials meaningfully—can be a form of cheap talk.
The problem is that one must often look beyond formal statements of the law to distinguish credible from noncredible assertions of state practice. Having made such an inquiry, we are skeptical of the view that absolute immunity was a default rule waivable by contract (see, e.g., Verdier and Voeten Reference Verdier and Voeten2015, 3 n4). Of course, we do not deny the existence of judicial opinions implying that sovereign immunity could be waived. We argue only that investors and governments likely viewed these as outlier cases that did not create a legal regime in which waivers would predictably be enforced. With regard to French law, for example, recall that not one of the bonds in our sample included a waiver of immunity despite the ostensibly clear rule giving effect to such clauses. The historical evidence indicates that investors were willing to try their luck at suing a foreign government, if they could persuade a court to take the case. So the answer does not seem to be that investors were indifferent to legal enforcement. The competing explanation—that investors viewed French courts as reluctant to assume jurisdiction over foreign sovereigns—suggests that legal experts may have assigned too much significance to the early French cases.
The broader point is that a robust inquiry into state practice requires more than a survey of traditional legal materials. When considering whether a judicial opinion, statute, or other formal statement of the law establishes state practice, it is appropriate to ask whether directly affected parties change their behavior in response. If this does not happen, the statement should not automatically be dismissed as cheap talk. (After all, a key insight in sociolegal studies is that parties do not always structure their affairs around legal rules.) Nevertheless, the red flag is up. When the statement of a rule has no effect on behavior, one potential explanation is that affected parties do not think the rule accurately predicts how state actors will behave. In such a case, the rule does not describe state practice, in any meaningful sense of that word.
A pessimistic response to our argument is to say that we propose to make the state practice requirement entirely unworkable. If CIL scholarship and practice already fails to take adequate account of traditional legal materials, what hope is there that jurists and scholars will undertake extensive empirical studies of market practice or anything else? However, we do not suggest that international tribunals should (or could) commission multiple empirical studies before ruling on contested matters. To be sure, a growing body of empirical work examines how state actors implement rules of international law (see, e.g., Whytock and Chilton Reference Whytock and Chilton2015; Choi and Gulati Reference Choi, Gulati and Bradley2016). But empirical studies will never conclusively resolve the doctrinal questions at the heart of CIL. We argue only that a wide range of evidence can shed light on whether state actors actually behave as implied by formal legal rules. Academics and practitioners should make a more concerted effort to look for such evidence, instead of confining themselves to an inquiry into traditional legal materials. In a sense, then, we call only for a more robustly empirical conception of CIL (see, e.g., Shaffer and Ginsburg Reference Shaffer and Tom2012).
A more robust conception of state practice, one that takes into account a wider range of empirical evidence, will not always yield easy answers. Indeed, such an approach will often complicate traditional conceptions of CIL rules. In a sense, then, one risk of expanding the focus beyond traditional legal materials is that this will reveal the limits of the very concept of state “practice.” We have little doubt that detailed empirical examination will often reveal state practices to vary in credibility, consistency, and generality. Where it exists, such variance makes the notion of state practice little more than an intellectual construct.
In other cases, however, a more robust empirical approach to identifying state practice will clarify rather than complicate the law on the books. Previously, for example, we discussed the uncertainty over whether pre-FSIA law in the United States treated waivers of sovereign immunity as irrevocable. Our inquiry into government bonds can help resolve this uncertainty. Recall that virtually all pre-FSIA government bonds issued in the US market omitted waivers of immunity. Because investors had reason to value the option to sue a sovereign that did not pay its debts, the likely explanation is that investors doubted the willingness of US courts (and the Department of State) to enforce such clauses.
Another benefit of looking beyond traditional legal materials is that such an inquiry can help to assess the relative significance of legal developments. Recall that conventional wisdom about the evolution of sovereign immunity law assigns relatively little significance to the FSIA and SIA. Yet when these statutes are judged by their impact on market practice, our data reveal them to be the most significant legal developments in the two-century history of the sovereign debt markets. The statutes transformed a global market and prompted market participants, for the first time, to incorporate legal enforcement rights into the contract template. By focusing only on traditional legal materials, students of sovereign immunity law have completely overlooked this radical transformation.
Conclusion
We began this project after noting a disjuncture in how our academic colleagues in the field of international law and our practitioner colleagues in the field of international finance understand the law of sovereign immunity. Most international law scholars describe a gradual, century-plus transition between two default rules. They understand sovereign immunity as a default rule and the absolute immunity regime as a baseline presumption of immunity that could be trumped by contract. By contrast, the practitioners appear to understand legal enforcement as a modern development, something that became possible only in the latter part of the twentieth century.
As scholars who study the design and evolution of contracts, we believe that market practices can shed light on this disagreement. If contracts are drafted in the shadow of legal rules, then these conflicting understandings of sovereign immunity law imply different predictions about contract design. If the law gradually shifted between default rules—from a presumption of immunity to a presumption of nonimmunity—then waivers of immunity should have been fairly common during the era of absolute immunity, when a creditor who had not obtained such a waiver would have no legal recourse. By contrast, if legal enforcement did not become relevant until late in the twentieth century, then contracts drafted earlier in the century should largely ignore the subject.
Our data largely support the practitioner view and imply that market participants viewed sovereign immunity as akin to a mandatory rule for much of the century. But the dramatic shift in contracting practices prompted by the FSIA and SIA calls for further explanation, as we can think of few legal changes that have had such a transformative impact on contract design. When market participants have strong incentives to coordinate, salient statutory enactments may shape behavior far more than other developments in the law.
