EU Bail-In Recognition Clause in U.S. Loan Agreements: An Empirical Review

I. Background

A.        The Bank Recovery and Resolution Directive 

            During the Great Financial Crisis of 2008, “EU member states were forced to use taxpayers’ money to prevent the failure of banks deemed too big to fail.”1 These bail-outs were politically controversial and created moral hazards encouraging future market participants to engage in excessive risk-taking.2 Indeed, the strain on the public purse was so significant that the government of Spain was left with a sovereign debt crisis that was at least partly the result of its bank bail-out efforts.3 In the aftermath of the European sovereign debt crisis, the nexus between sovereign debts and bank bail-outs attracted widespread public and political attention.4

            In June 2014, the EU finalized and adopted the Bank Recovery and Resolution Directive (2014/59/EU) (BRRD) “creating a harmonised framework across Europe for dealing with the problem of ‘too big to fail’ through bank recovery and resolution.”5 The BRRD is a minimum harmonization directive requiring member states to appoint a resolution authority, usually their central bank or existing bank regulator, and, among other things, to give such resolution authority the power to write down unsecured liabilities or convert them into equity.6 Such write-down and conversion power is typically described as the power to bail-in.7 In contrast to bail-out, which uses public funds to recapitalize an ailing systematically important bank, bail-in uses funds from the bank’s creditors to recapitalize by extinguishing their claims.8 Therefore, the BRRD’s bail-in power transfers at least the initial burden of rescuing banks from taxpayers to the bank’s own creditor investors.9

B.        Article 55

            Naturally, the BRRD provides for automatic intra-EU recognition of any national resolution authority’s exercise of write-down and conversion powers; therefore, as long as an agreement evidencing or creating liabilities is governed under the laws of an EU member state, bail-in actions are enforceable as a matter of law.10 However, as EU law and any powers exercised thereunder do not automatically apply extra-territorially, there is a risk that bail-in actions cannot effect liabilities arising from contracts governed under non-EU law if the courts of that non-EU jurisdiction refuse to recognize the relevant EU resolution authority’s write-down and conversion powers.11

            To mitigate this risk, Article 55 of the BRRD requires EU banks and other in-scope financial firms to include a contractual recognition of EU national authorities’ bail-in power in every non-EU law governed agreement that might give rise to liabilities and is able to be bailed-in under the BRRD.12 “The rationale behind these recognition clauses is that counterparties are less likely to successfully challenge the effectiveness of resolution actions if they contractually agree to be subject to such actions.”13 Specifically, Article 55 requires the recognition clause to include a description of the bail-in power, the contractual counterparties’ acknowledgement and acceptance that the in-scope EU entity’s liability may be subject to the exercise of write-down and conversion powers by a resolution authority, and the counterparties’ agreement to be bound by any decisions to write down or convert.14 However, the BRRD does not provide a specific form of the contractual clause.15

            Under Article 55, it is very clear that a bail-in recognition clause must be included in all debt instruments, such as bond indentures, that an in-scope entity issues as a borrower. However, since the BRRD defines liability broadly to include contingent liabilities in any form unless explicitly excluded, Article 55 also requires an in-scope entity to include a bail-in recognition clause even in contracts where it acts as a lender.16 A loan facility “constitutes an asset of the lending bank rather than a liability but Article 55 will still be relevant to a bank’s commitment to extend credit under a loan and its potential liabilities to other syndicate members in a loan or an inter-creditor arrangement.”17

            U.S. law governed credit agreements fall under the purview of Article 55 if an in-scope EU financial institution is a lender under the agreement or will become a member of the syndicate through future purchases of participation interests on the secondary market.18 As EU-based financial institutions are a major part of the U.S. primary and secondary syndicated loan markets, the need to add or include bail-in recognition clauses quickly caught the attention of U.S. loan market participants before the Article 55 requirements became effective on January 1, 2016.19 For example, the Loan Syndications and Trading Association (LSTA), the premier industry organization for debt financing in the U.S., issued a model provision to be used in New York law governed loan documents in December 2015.20 Additionally, many Wall Street law firms issued client notices and advisories ahead of Article 55’s implementation.21

            It has been a full decade since Article 55 came into force, but aside from one early cursory study in 2016, there has been virtually no empirical research examining bail-in recognition clauses in U.S. law governed credit agreements.22 This study aims to fill this void by assembling a proprietary dataset to analyze the prevalence, drafting, and impact of Article 55 bail-in recognition clauses in U.S. law governed credit agreements.

II. Motivation & Hypotheses

            When Article 55 became effective in 2016, most U.S. market participants expected a bail-in recognition clause to be included in the vast majority of large corporate loan documentations regardless of whether an in-scope EU financial institution is an original party to the agreements or not.23 These loans are often broadly syndicated and each lender’s commitment can be traded or subdivided through selling participations on the secondary market; as a result, any in-scope EU financial institution could become a lender under these loan agreements even after primary syndication.24 Therefore, the absence of a pre-baked bail-in recognition clause will limit the pool of eligible purchasers in secondary loan trading because in-scope EU financial institutions will not be able to buy into a loan without amending the underlying loan documentation. To date, however, there has been no empirical analysis validating bail-in clauses’ prevalence in U.S. law governed credit agreements.

            As noted in Section I.B, the LSTA issued a model bail-in recognition provision to be used in New York law governed credit agreements. I expect the bail-in recognition clause contained in most SEC-filed credit agreements to be nearly identical to the LSTA model provision (H1) because the potential costs of deviation is high and the transaction costs of drafting bespoke formulations is also high. The consequences of breaching the Article 55 requirement vary depending on individual member state’s implementing legislation, but it may include public censure and a potentially unlimited fine.25 The EU and its national resolution authorities have implicitly endorsed the LSTA formulation as compliant with Article 55 as it has never found it to be non-compliant in the past decade. Therefore, any deviation from the LSTA formulation comes with a risk of non-compliance. Additionally, prior research has established that the debt finance industry is prone to treat terms as boilerplate and avoid expending transaction costs on drafting bespoke terms because both sides of a deal are driven primarily by short-term incentives.26

            I also expect that it is less likely for credit agreements in non-bank-intermediated lending, also known as private credit, to contain bail-in recognition clauses than traditional bank-intermediated credit agreements (H2). As discussed above, bank-intermediated corporate lending necessitates the incorporation of bail-in provision because these loans are often widely syndicated and traded in a large secondary market that may ultimately involve an in-scope EU financial institution. In contrast, most private credit lenders, whether they are alternative asset managers or insurance companies, fulfill their lending commitments with existing funds under their management rather than syndicating or selling participations to other lenders.27 In other words, they have greater ex ante control over whether or not any in-scope EU financial institutions will end up involved in a transaction. If they determine that no in-scope EU financial institutions will ever become lenders under a credit agreement, they may exclude the bail-in recognition clause.

III. Data Collection

            This study uses a proprietary dataset I assembled that covers all U.S. law governed credit agreements entered into during the 6-month period between January 1, 2025, and June 30, 2025, that are publicly filed with the SEC. These credit agreements were filed pursuant to 17 C.F.R. § 229.601, which requires public companies to file all material contracts as Exhibit 10 to their Form 10-K and Form 10-Q filings.28 The dataset contains 242 observations and includes the names of the borrowers and administrative agents as well as three study variables and three control variables. Summary statistics are presented in Table 1.

            Non-Bank Admin Agent is coded as a binary study variable that equals 1 when the administrative agent is a private credit lender, and equals 0 when the administrative agent is a bank. The private credit lender classification is a residual category covering all entities that are not a federal, state, or foreign chartered bank. In this dataset, private credit lenders include alternative asset managers, insurance companies, acquirers in acquisition financing, and ad hoc lenders that do not regularly participate in financing activities.

            Bail-In Clause is also coded as a binary study variable that equals 1 when the credit agreement contains a clause recognizing EU resolution authorities’ write-down and conversion power under the BRRD, and equals 0 if not.

            WCopyFind Score is a continuous study variable with a maximum value of 100 and a minimum value of 0. If a credit agreement contains a bail-in recognition provision, I used WCopyFind 4.1.5, a textual similarity software, to compare the operative text of such provision to the LSTA model provision.29 WCopyFind is originally created to detect plagiarism by university students but it has also been used to analyze the extent to which U.S. Supreme Court opinions copy from lower court decisions, docket filings, and amicus briefs.30 In using the software, I followed conventions set forth in prior research to set the shortest strings of words to six but departed from conventions to set the other parameters to the most restrictive level possible; this ensures that only true word-for-word exact matches are factored into the similarity scores.31 Therefore, a WCopyFind Score of 100 means the bail-in recognition provision is 100% identical to the LSTA model provision.

            Additionally, I included three other variables as controls. First, the variable maturity reflects the tenor of loans extended under the credit agreement; if there is more than one type or tranche of loan extended with varying tenors, maturity reflects the longest tenor. Second, the variable value reflects the maximum total lending commitments under the credit agreement. Third, the variable secured reflects whether or not the loans are secured or unsecured.

IV. Methodology & Results

A.        Descriptive Analysis 

            Of the 242 credit agreements analyzed, 224 of them contained a bail-in recognition provision while 18 did not. In other words, 92.56% of the sample analyzed contained a bail-in recognition provision. Although the dataset is limited to a 6-month period in 2025, this result should be generalizable as it is derived from a reasonably large sample (N = 242) and there is no reason to expect temporal variation given that the BRRD Article 55 requirements have remained unchanged since its inception in 2016. Therefore, early market participants correctly expected that bail-in recognition clauses would become prevalent in large corporate credit agreements governed under U.S. law.

B.        Hypothesis 1: Bail-in clauses in SEC-filed credit agreements are identical to the LSTA model provision

            The median similarity score for bail-in clauses in the dataset is 97.5% with a mean of 94.81%. Table 2 below provides the mean and five-number summary of the WCopyFind Score variable. Given that I applied the most restrictive parameters in generating these similarity scores, a score around 95% indicates identity in all material respects with only minor conforming changes; for example, observation #13 differs from the LSTA model only to the extent that it refers to the auxiliary documents as “Credit Documents” rather than “Loan Documents” and observation #241 differs only to the extent that it adds a reference to “Issuing Banks” given that the deal involves letter of credit facilities.

            Indeed, only about one fifth of the dataset has a WCopyFind score below 90%. For nearly all of these observations, their variation from the LSTA model provision can be attributed to one of two additions to the LSTA model. First, some observations contain an additional notice requirement at the end of the clause that mandates lenders to notify the borrower and the administrative agent if it becomes an in-scope financial institution under any EU resolution authorities (e.g., they are

headquartered in the U.S. when the credit agreement is signed but subsequently redomesticate to an EU jurisdiction).32 Second, some observations contain an additional sentence at the beginning of the clause making explicit the self-evident proposition that any bail-in actions will only eliminate the commitments of those lenders who are bailed-in but not other lenders’ commitments.33 Besides having one of these two additions, this minority of observations with scores below 90% share all key operative language with the LSTA model.

            Nevertheless, there is one extreme outlier with a similarity score of 0%. Interestingly, however, the clause contained in this observation appears to be identical to the Loan Market Association’s (LMA) model provision. The LMA is the English law counterpart of the LSTA, which also publishes model provisions to be used for credit agreements governed under English law.34 I regard this outlier as the result of an inadvertent mistake made in choosing precedents at a trans-Atlantic firm. Indeed, the fact that the software returned a 0% similarity score even though the LMA and LSTA model provisions use largely the same words but in different order demonstrates the robustness of the software in only detecting real similarity in strings of six words. Therefore, with minor variations as discussed above, H1 is confirmed.

C.        Hypothesis 2: It is less likely for private credit deals to include a bail-in recognition clause

            As the outcome variable Bail-In Clause is binary and the study explanatory variable Non-Bank Admin Agent is also binary, I estimated using a logistic regression model with the following specification to avoid heteroskedasticity and risk of boundary violations from a linear probability model35:

            Table 3 below shows the results of this logistic regression model. The estimated coefficient for Non-Bank Admin Agent is –1.639 (p = 0.0175), which is statistically significant at the 5% level. If we convert this estimate to an odds ratio, this means that private credit deals are 80.6% less likely to contain a bail-in recognition clause than bank-intermediated deals.36

              As private credit lenders generally engage in a higher proportion of lower value middle-market deals, I was concerned that there is a high correlation between Non-Bank Admin Agent and the control variable Value.37 I therefore tested for multicollinearity using the variance inflation factor test and found no problematic correlation between any of the explanatory variables. Therefore, with a statistically significant estimate that private credit deals are 80.6% less likely to contain a bail-in recognition clause than bank-intermediated deals, H2 is confirmed.

V. Discussion

            When Article 55 came into effect in 2016, market participants predicted that a bail-in recognition clause would be included in the vast majority of large corporate credit agreements governed under U.S. law.38 This study empirically validates that prediction, showing that 92.56% of SEC-filed credit agreements contain such a clause as mandated by Article 55. The prevalence of contractual clauses in U.S. law governed credit agreements recognizing EU member states’ power to write-down and convert liability demonstrates that the EU can exert influence over market behavior beyond its territorial boundaries simply through regulating domestic EU actors. De jure, the EU cannot force U.S. borrowers or administrative agents to adopt a bail-in clause or any other contractual clauses, but the EU has de facto power to do so because the cost of completely cutting off European lenders is too high for the U.S. loan market to bear.

            Although bail-in recognition clauses are now omnipresent in U.S. law governed credit agreements, whether they, and by extension the EU’s underlying bail-in actions, will be enforceable under New York law or other state law is still an open question as they have not been tested in court.39 The recognition of EU bail-in actions raises domestic public policy concerns in non-EU jurisdictions as it allows foreign regulatory actions to directly affect and derogate the rights of domestic borrowers and co-lenders under agreements that are governed under its domestic law. On this issue, scholars and practitioners in Switzerland have questioned whether specific formulations of EU bail-in recognition clauses are enforceable under Swiss law given that they affect domestic legal entitlements and alter domestic debt restructuring laws.40

              Even though there is a presumption in U.S. law that contractual terms between sophisticated parties are enforceable, courts might heavily scrutinize and narrowly construe the reach of these clauses. Indeed, in observation #20, the opinion letter of Citibank’s special New York counsel explicitly states that they “express no opinion as to whether inclusion of the bail-in clause in Section 8.17 of the Credit Agreement or any Bail-In Action under it will be given effect.”41 Likewise, the opinion letter of Brown Rudnick LLP as borrower’s counsel in observation #20 refuses to render an opinion on “the enforceability of any ‘bail in’ clause or similar provision.”42 Therefore, the fact that nearly all EU bail-in recognition clauses in U.S. law governed credit agreements are operatively identical to the LSTA model provision presents a systematic risk to the BRRD’s goal of ensuring its resolution authorities’ bail-in actions are recognized and enforced in the U.S. For instance, a single adverse ruling from a New York court invalidating or refusing to enforce the LSTA’s formulation will invalidate bail-in clauses in almost every New York law governed credit agreement as they are identical to the LSTA model. Although EU authorities have implicitly accepted that the LSTA formulation satisfies Article 55’s requirements, it remains an open question whether New York courts will accept this particular formulation that is now boilerplate.43 The pari passu saga in the sovereign bond realm teaches us that boilerplate clauses that are widely and mindlessly copied from precedents or models may pose systematic risks when their interpretation or enforceability are completely untested in court.44

              Lastly, the finding that private credit deals are 80.6% less likely than bank-intermediated deals to contain a bail-in recognition clause is further evidence that private credit lenders have greater flexibility in the loan market than bank lenders.45 Intuitively, not being potentially subject to EU bail-in actions is at least somewhat valuable to a corporate borrower; ceteris paribus, a borrower would prefer a private credit lender whose lending commitments are unqualified over a bank-intermediated loan where lending commitments can be written down through EU regulatory action. Indeed, the European Association of Co-operative Banks have warned that the Article 55 contractual recognition clause requirement imposes a competitive disadvantage on European bank lenders.46 The past decade of Article 55’s existence coincided with a massive boom in private credit both in terms of dollar volume and market share.47 While I am certainly not suggesting that Article 55 was a major contributing factor to private credit’s meteoric rise, the ability to bypass Article 55 is yet another dimension evidencing private credit’s flexibility and agility vis-à-vis traditional bank-intermediated lending.

VI. Conclusion

Almost a decade after BRRD Article 55 came into force, research on its effect and implementation in the U.S. loan market remains almost entirely nonexistent. Through collecting a proprietary dataset of SEC-filed credit agreements, this study makes three empirical contributions. First, I show that EU bail-in recognition clauses are ubiquitous in U.S. law governed credit agreements, at least for large corporate loans. This reflects yet another dimension where the EU is able to exert extra-territorial influence. Second, I present evidence showing that bail-in clauses in nearly all observations are operatively identical to the LSTA model provision. This presents a systematic risk to BRRD’s effectiveness as the judicial invalidation of the bail-in clause in any of these agreements will ipso facto nullify bail-in actions vis-à-vis almost all U.S. loan agreements. Third, the results also indicate that it is 80.6% less likely for private credit deals to include a bail-in recognition clause. This adds to the private credit literature’s well-developed understanding that private credit provides borrowers with greater flexibility than traditional bank-intermediated lending.

• Practical Law Financial Services, Bank Recovery and Resolution Directive (BRRD), Thomson Reuters Practical Law (last visited Nov. 17, 2025), https://uk.practicallaw.thomsonreuters.com/1-576-2705 [https://perma.cc/KF3K-PQJS].
• Alexander Schäfer, Isabel Schnabel & Beatrice Weder di Mauro, Bail-In Expectations for European Banks: Actions Speak Louder than Words 2 (ESRB Working Paper Series No. 2016/07), https://ssrn.com/abstract=3723353 [https://perma.cc/S3NR-MPDN].
• See Jeffry Frieden & Stefanie Walter, Understanding the Political Economy of the Eurozone Crisis, 20 Annu. Rev. Polit. Sci. 371, 377 (2017).
• Livia Pancotto, Owain ap Gwilym & Jonathan Williams, The European Bank Recovery and Resolution Directive: A Market Assessment, 44 J. Fin. Stab. 1, 1 (2019).
• Bank of England Prudential Regulation Authority, Consultation Paper: Implementing the Bank Recovery and Resolution Directive 5 (July 2014).
• See Practical Law Financial Services, supra note 1.
• Jannic Alexander Cutura, Debt Holder Monitoring and Implicit Guarantees: Did the BRRD Improve Market Discipline?, 54 J. Fin. Stab. 1, 1 (2021).
• Benjamin Bernard, Agostino Capponi & Joseph E. Stiglitz, Bail-Ins and Bailouts: Incentives, Connectivity, and Systemic Stability, 130 J. Polit. Econ. 1805, 1806–07 (2022).
• David Marques-Ibanez, Gianluca Santilli & Giulia Scardozzi, Bail-In in Action 3 (European Central Bank Working Paper No. 2024/2959), https://ssrn.com/abstract=4907318 [https://perma.cc/42KJ-FL7X].
• Silvia Morlino, Contractual Recognition of Bail-in, 11 Int’l In-House Couns. J. 1, 2 (2018).
• Id. 
• Clifford Chance LLP, Recognition of EU bail-in clauses – key considerations for the Asia Pacific market 1 (Oct. 2015).
• Morlino, supra note 10, at 2.
• Addleshaw Goddard LLP, New Article 55 BRRD Bail-In Waiver 4 (2015).
• Morlino, supra note 10, at 6–7.
• See Barbara M. Goodstein, Bail-Out Turns to Bail-In: Europe Anticipates Next Financial Crisis, 255 N.Y. L.J. 66, 1 (Apr. 7, 2016).
• Clifford Chance LLP, supra note 12, at 3.
• Id. at 4.
• Goodstein, supra note 16, at 2.
• Id.
• E.g., Claude Brown, Elizabeth McGovern & Colin Cochrane, Bail Out the Sea of Paper in Your In-Box – Understanding Article 55 Bail-In Clauses, Reed Smith LLP Client Alerts (Jan. 26, 2016), https://www.reedsmith.com/en/perspectives/2016/01/bail-out-the-sea-of-paper-in-your-inbox--understan [https://perma.cc/R4LT-GCN7]; Nick Shiren & Assia Damianova, Contractual Recognition of Bail-In – Are You Ready?, Cadwalader, Wickersham & Taft LLP Resources: Clients & Friends (Dec. 15, 2015), https://www.cadwalader.com/resources/clients-friends-memos/contractual-recognition-of-bail-in-are-you-ready [https://perma.cc/V8W4-VJQ9].
• See Practical Law Finance, What's Market: EU Bail-In Rules, Thomson Reuters Practical Law (last visited Nov. 17, 2025), https://us.practicallaw.thomsonreuters.com/w-003-6629 [https://perma.cc/9DJR-QNMX].
• Jane Summers, Alan W. Avery & Alfred Y. Xue, Latham & Watkins Explains US Loan Market Adaptations to European Bail-In Directive, Columbia Law School Blue Sky Blog (Feb. 8, 2016), https://clsbluesky.law.columbia.edu/2016/02/08/latham-watkins-explains-us-loan-market-adaptations-to-european-bail-in-directive/ [https://perma.cc/4AHT-SXYF].
• Id. 
• Clifford Chance LLP, supra note 12, at 4.
• Stephen J. Choi, Robert E. Scott & G. Mitu Gulati, Revising Boilerplate: A Comparison of Private and Public Company Transactions, 2020 Wis. L. Rev. 629, 630 (2020).
• Peter Pollini & Daniel Sullivan, Private Credit: The Rewiring of Credit in Capital Markets, PricewaterhouseCoopers LLP Reports (May 29, 2025), https://www.pwc.com/us/en/industries/financial-services/library/private-credit.html [https://perma.cc/94UL-9XWG].
• 17 C.F.R. § 229.601 (2019).
• Headings are excluded because most credit agreements stipulate that headings are for convenience of reference only and do not control, change, or limit the meaning or interpretation of any provisions.
• Evelyne Brie, Cynthia Huo & Christopher Alcantara, Measuring Policy Diffusion in Federal Systems: The Case of Legalizing Cannabis in Canada under Time Constraints, 54 Publius: The J. of Federalism 228, 238–39 (2023); see also, e.g., Pamela Corley, The Supreme Court and Opinion Content: The Influence of Parties’ Briefs, 61 Pol. Res. Q. 486 (2008); Adam Feldman, Opinion Construction in the Roberts Court, 39 L. & Pol’y 192 (2016).
• Brie et al., supra note 30, at 239.
• See Observation #46.
• See Observation #18.
• Cadwalader, Wickersham & Taft LLP, LMA vs. LSTA loan trading 3 (Dec. 2015).
• Robert L. Kaufman, Heteroskedasticity in Regression: Detection and Correction 16 (2013).
• As the model follows a log-odds specification, we convert by calculating = 0.194. Therefore, the odds of a private credit deal having a bail-in recognition clause is only 19.4% of the odds of a bank-intermediated deal having such a clause.
• Frank Oliver & Michelle L. Iodice, Overview and Comparison of the Broadly Syndicated Loan And Private Credit Markets, Proskauer Rose LLP Global Legal Insights (Nov. 12, 2025),

  https://www.proskauer.com/pub/overview-and-comparison-of-the-broadly-syndicated-loan-and-private-credit-markets [https://perma.cc/6GYK-HBRB].

• Summers et al., supra note 23.
• Goodstein, supra note 16, at 2.
• Rashid Bahar et al., Bail-in Recognition Clause, 5 Swiss Cap. Mkt. L. 7, 16–17 (2016).
• Exhibit C–2 to 364-Day Credit Agreement dated as of June 23, 2025, between Stanley Black

  & Decker Inc., and other lenders thereto (on file with the Securities and Exchange Commission

  https://www.sec.gov/Archives/edgar/data/93556/000119312525148898/d810765dex101.htm [https://perma.cc/4R86-Z3A8]).

• Exhibit C–1 to 364-Day Credit Agreement dated as of June 23, 2025, between Stanley Black

  & Decker Inc., and other lenders thereto (on file with the Securities and Exchange Commission

  https://www.sec.gov/Archives/edgar/data/93556/000119312525148898/d810765dex101.htm [https://perma.cc/R477-XGY4]).

• Goodstein, supra note 16, at 2.
• See generally Rodrigo Olivares-Caminal, The Pari Passu Clause in Sovereign Debt Instruments: Developments in Recent Litigation, BIS Q. Rev. Paper No. 72 (Feb. 20, 2013).
• See Practical Law Finance, supra note 22.
• European Association of Co-operative Banks, EACB Response to EBA Consultation on Technical

  Standards on Impracticability of Contractual Recognition of Bail-In, EACB Position Papers (Oct. 23, 2020), https://www.eacb.coop/en/recovery-resolution-deposit-protection/position-papers/eacb-response-to-eba-consultation-on-technical-standards-on-impracticability-of-contractual-recognition-of-bail-in.html [https://perma.cc/5XL7-62JB].

• Jared A. Ellias & Elisabeth De Fontenay, The Credit Markets Go Dark, 134 Yale L. J. 779, 785 (2025).