The Limits of Enforcement in Global Financial Governance: Blacklisting in FATF as Rational Myth

Abstract

How might international institutions matter? To consider this central question of International Relations, we analyze a most-likely case for the importance of materially driven enforcement: the Financial Action Task Force’s (FATF) use of blacklisting in the global regime targeting money laundering and terrorism financing. Scholars and practitioners often argue that fear of financial harm caused by FATF’s lists explains the near-global commitment to FATF’s standards, even if compliance lags. We search for statistical evidence of this impact across four different measures of financial flows and find that listing is not correlated with financial harm. To explain these null results, we examine bank decision-making and find that the lists’ impact is likely diminished by two overlooked factors: the existence of multiple, competing lists and banks’ access to more fine-grained, client-specific information provided by third-party companies. We interpret this contradiction—a commitment to compliance generated in part by a fear of enforcement, despite a lack of evidence for enforcement’s impact—as a “rational myth.” The results challenge a common understanding of a major global governance regime, confirm ideas about the limited ability of states or International Organizations to control governance outcomes, and advance a new research agenda on the impact of bank decision-making on global governance.

¿Qué importancia podrían tener las instituciones internacionales? Analizamos, con el fin de poder considerar esta cuestión tan relevante para las Relaciones Internacionales, uno de los casos más probables que existen a favor de la aplicación de la ley impulsada materialmente: el uso por parte del Grupo de Acción Financiera Internacional (GAFI) de la inclusión en listas negras en el régimen global contra el lavado de dinero y el financiamiento del terrorismo. Los académicos y los profesionales argumentan, con frecuencia, que el temor al daño financiero causado por las listas del GAFI explica la existencia de un compromiso, casi global, con los estándares del GAFI, incluso si el cumplimiento se retrasa. Buscamos evidencia estadística de este impacto a través de cuatro medidas diferentes de flujos financieros y concluimos que el hecho de figurar en la lista no está correlacionado con ningún perjuicio a nivel financiero. Con el fin de poder explicar estos resultados nulos, estudiamos la toma de decisiones bancarias y concluimos que el impacto que ejercen estas listas se ve disminuido debido a dos factores que normalmente no se tienen en cuenta: la existencia de múltiples listas que compiten entre sí y el acceso por parte de los bancos a información más detallada y específica del cliente, la cual es proporcionada por terceras empresas. Interpretamos esta contradicción (el compromiso con respecto al cumplimiento generado, en parte, por el miedo a la aplicación de la ley, a pesar de la falta de pruebas sobre el impacto de la aplicación de la ley) como un “mito racional”. Nuestros resultados desafían la comprensión común relativa a un régimen de gobernanza global importante, confirman las ideas sobre la capacidad limitada de los Estados o las Organizaciones Internacionales para controlar los resultados de la gobernanza y promueven una nueva agenda de investigación sobre el impacto de la toma de decisiones bancarias en la gobernanza global.

Quelle est l'importance des institutions internationales ? Pour traiter cette question centrale dans les RI, nous analysons un cas fort probable d'application aux motivations financières : l'utilisation de la mise sur liste noire par le groupe d'action financière (GAFI) dans le régime mondial par rapport au blanchiment d'argent et au financement du terrorisme. Les chercheurs et les professionnels affirment souvent que la peur des répercussions financières dues aux listes du GAFI explique le fait que quasiment le monde entier s'engage à respecter ses normes, même si la conformité tarde. Nous recherchons des éléments statistiques pour venir étayer cette incidence dans quatre mesures de flux financiers pour conclure à l'absence de corrélation entre l'inscription sur une liste et des conséquences financières négatives. Pour expliquer ces résultats nuls, nous examinons la prise de décisions des banques et observons que deux facteurs négligés atténuent probablement l'effet des listes : l'existence de nombreuses listes concurrentes et l'accès des banques à des informations plus détaillées et spécifiques aux clients par le biais d'entreprises tierces. Nous interprétons cette contradiction—un engagement à la conformité générée en partie par une peur de l'application, malgré une absence de preuves de l'incidence de l'application—tel un « mythe rationnel ». Les résultats remettent en question une conception courante du régime de gouvernance mondial prédominant, confirment les idées quant aux capacités limitées des États ou des OI quand il s'agit de contrôler les issues de la gouvernance, et promeuvent un nouveau programme de recherche sur l'incidence de la prise de décisions des banques sur la gouvernance mondiale.

Introduction

Whether, when, and how international institutions matter remain central questions in International Relations.¹ We contribute to this debate by analyzing the impact of the Financial Action Task Force (FATF), a governance network at the center of the international anti-money laundering and counter-financing of terrorism (AML/CFT) regime. Over the last 35 years, the AML/CFT regime has evolved from an obscure gathering of experts (Strange 1998) into a global standard.

Academics, financial experts, and the media often attribute FATF’s influence to its use of public lists that identify states in various “shades” of non-compliance: a “blacklist” for the worst offenders and a “greylist” for supposedly less egregious non-compliers. We refer to this listing process as a whole as “blacklisting” or “listing” and the lists together as “the lists.” Many observers argue that the power of the lists comes via what Morse (2019, 2022) terms “unofficial market enforcement”: being listed drives away investors, especially those based in the wealthiest countries. To avoid this economic harm, states commit to comply with the AML/CFT regime.

This expectation is intuitive. Key FATF supporters, e.g., the United States, the United Kingdom, and EU, hold an overwhelming preponderance of financial power. The lists also seem to be designed to facilitate material enforcement. In practice, however, the evidence regarding blacklisting’s financial impact is surprisingly mixed. So what impact does blacklisting have and what does that suggest about the impact of international institutions?

Based on a statistical evaluation of financial flows and process tracing of bank decision-making, we argue that the best available evidence points to a belief in the financial impact of blacklisting, but also to a lack of actual financial harm. While the belief defies empirical evidence, it nonetheless generates action from otherwise reluctant states. We interpret this dynamic as a “rational myth,” a framework that highlights how collective expectations of rational design shape institutions’ actions and participants’ assessment of institutional performance (Meyer and Rowan 1977). Institutions adopt rationalized procedures, but do so “ceremoniously” and with little regard for actual impact (Meyer and Rowan 1977, 340). Observers accept those procedures and assume that outcomes match expectations, due in large part to the “rationalism, formalism, and intellectual rigour” that characterize rationalized institutions (Boiral 2007, 128).

FATF’s lists and the supposedly technical process behind them construct the AML/CFT regime as a rational, effective governance structure that provides objective, useful information to states and private actors who respond accordingly. The acceptance of that myth leads otherwise reluctant states to change their regulatory systems, further reinforcing the myth. However, implementation and compliance are also largely ceremonial. FATF adopts blacklisting, but does not track its deeper impact. States agree to FATF’s standards, but fail to ensure that banks implement the rules. Even banks voice support for the rules, but fail to ensure that changes in routines lead to changes in outcomes.

We lay out this argument in three fundamental steps. First, we triangulate various sources to establish that observers expect the lists to harm targeted economies. Second, we look for evidence of that impact. Following Easterly’s (2003) example, we re-consider the strongest evidence to date supporting the hypothesis: statistical analysis showing a negative correlation between a country’s listing and cross-border banking liabilities. We use a more complete dataset of FATF blacklisting and augment the model to integrate recent advances in the use of two-way fixed effects (TWFE) models in samples with time-varying treatments. The results yield no statistically significant correlation between being listed by FATF and four different measures of financial flows: cross-border banking liabilities, portfolio asset flows, foreign direct investment (FDI), and World Bank development assistance. These multiple null results suggest that listing does not create the widespread negative impact that many observers expect.

Third, to make sense of those null results, we use existing research and qualitative process tracing to analyze investment decisions in banks, which is where blacklisting should change outcomes. Two factors stand out, both largely ignored in existing scholarship. One is the prominence of lists as tools of governance (Amicelle and Jacobsen 2016). The FATF lists are just two of tens or even hundreds of different, often conflicting (Riccardi 2022) lists that banks are supposed to use to filter clients. A second factor is the common use of third-party “Know Your Customer” (KYC) companies that provide banks with real-time, individualized risk profiles of clients. In comparison, FATF lists provide vague, low-quality information. Given the various and conflicting signals from different regulatory lists and from KYC companies, it would be puzzling if FATF blacklists guided bank decisions as many expect. Yet actors within the regime routinely act as if they did.

These findings hold several implications for ongoing scholarship. They suggest we need a deeper understanding of the various mechanisms behind the AML/CFT regime’s expansion. The research advances a newly established research agenda on lists as governance and the role of “KYC” companies in financial relations (de Goede and Sullivan 2016). It also advances work on the role of rational myths in global financial governance. It suggests an important limit on states’ abilities to “weaponize” global financial interdependence (Farrell and Newman 2019) and on IOs’ ability to “orchestrate” global public policy (Findley, Nielson, and Sharman 2015b; Abbott et al. 2021). More broadly, the findings confirm recent advances showing the defining role of ideas in the global political economy. Methodologically, the research demonstrates recent advances in the use of TWFE models with time-varying treatments.

In the next section, we briefly describe FATF and the listing process before placing it within broader debates over compliance. In section 2, we discuss the mixed-method research design and strategy. Section 3 presents our empirical results: the expectation of a financial impact from FATF’s lists, the lack of evidence for that impact, and an explanation for those null results based on bank decision-making. In the final section, we interpret the results as an instance of a rational myth and discuss the broader implications of the findings.

Understanding the FATF Lists

In 1989, the eleven founding members of FATF adopted a 1-year mandate to identify common practices across states fighting profits from the drug trade. Those efforts eventually resulted in the “FATF 40 Recommendations” for building stronger AML systems. FATF’s portfolio has since expanded to address transnational organized crime, terrorism financing, and proliferation financing. Current members have halted membership at 37, but most countries and territories are members of at least one of the nine regional iterations of FATF, the “FATF-Style Regional Bodies,” and have committed to meeting FATF’s standards. There are twenty-eight observer bodies, including the International Monetary Fund and various organs of the World Bank and United Nations. FATF is a transnational public policy network (Reinicke 1999; Slaughter 2009) with remarkable influence for a “voluntary” initiative.

Early on, FATF members rejected a reliance on enforcement (Hülsse 2008) and opted instead for tools that would generate new knowledge and create peer pressure: peer reviews, mutual evaluations, and “typologies exercises” to identify new trends in money laundering. The archival record shows that members explicitly rejected the use of blacklists (Nance 2018). Members first aimed to enforce the 40 Recommendations in 2000, 11 years after FATF’s founding. They have consistently revised the listing process, suggesting that the system was not working as members had intended (Nance 2018).

That said, the blacklisting process that has been in place since 2010 now seems designed to threaten. Relatively invasive “mutual evaluations” occur roughly every 10 years. Teams of experts, often seconded from other international organizations, review countries based on a nearly 200-page handbook known as the Common Methodology. A technical review of laws sets the focus for the more in-depth review carried out in an on-site visit. After the visit, reviewers generate a draft report, which the review committee and the reviewed jurisdiction discuss and sometimes revise. Members discuss the report in the FATF plenary before publishing a final version. Members officially decide whether to list a jurisdiction based on the results of this review process.

FATF then issues two statements that comprise the lists. The “Public Statement” indicates jurisdictions with serious deficiencies and further distinguishes between two groups. The most serious category of the Public Statement calls for states to enforce countermeasures against the listed jurisdictions; states should require financial institutions to restrict or eliminate financial transactions with entities in the listed jurisdiction. Only Iran and North Korea have ever been on that true blacklist. The second category in the Public Statement indicates countries that are deficient but not facing countermeasures. A rotating cast of large and small developing countries and jurisdictions has been on the list: fifty-two in total since 2010.²

The second statement FATF issues is entitled “Improving Global AML/CFT Compliance: On-going Process.” It names jurisdictions that have deficiencies but also have made a “high-level political commitment” to address them. These jurisdictions are subject to increased monitoring. While the lists denote three shades of non-compliance, as portrayed in Table 1 below, in practice, the public discussion is less nuanced and shifts across outlets and over time. Table 1 below represents these groupings.

Among other obligations, compliant states require banks under their supervision to conduct “KYC” assessments in order to verify “beneficial ownership,” or the person who ultimately owns the account. Banks must report any suspicious transactions to the appropriate domestic authorities. Most importantly, banks must be required to subject transactions with entities in listed jurisdictions to additional scrutiny or to eliminate them entirely. The regime depends heavily on the work of these agents (Sharman 2009). Ultimately, the AML/CFT regime deputizes banks and other financial institutions to apply national interpretations of international standards. The central question here is whether this chain of delegation leads to the financial harm that many observers expect.

“Getting to Yes”: IR Debates over Compliance

FATF and the AML/CFT regime present an interesting laboratory for testing theories about transnational cooperation and the role of enforcement. The regime generates a “simultaneous financialization of security and securitization of finance” (de Goede 2020), creating a deeply significant regime complex (Keohane and Victor 2011; Margulis 2013). Debates across IR, sociology, and finance are relevant and provide competing explanations.

Within IR, some scholars argue that “deep cooperation,” or action that would not happen absent an agreement, requires strong enforcement to allay fears of asymmetric losses. Enforcement is also subject to the collective action problem, however, so having a small group of actors with the preponderance of power is likely necessary for deep cooperation (Downs, Rocke, and Barsoom 1996). While these requirements are stringent, FATF would seem to fulfill them.

Others argue against this primary role for enforcement. These “managerialists” argue that compliance is the norm and that non-compliance is rare, often unintentional, or the result of good-faith disagreements or genuine lack of capacity (Chayes and Chayes 1991, 1993). And while there may be a role for enforcement, enforcement requires agreement on what constitutes a violation, which is rare (Koh 1996). Improved compliance thus requires mechanisms that could promote common understandings of obligations, arbitrate in cases of disagreement, and facilitate capacity building (Chayes and Chayes 1990; Checkel 2001). FATF has several tools designed to elaborate and improve upon common understandings of its standards, which point toward compliance with different motivations.

Scholars also debate the effect of actor proliferation on cooperation, which is germane to FATF. For some, the proliferation of actors within global governance exacerbates collective action problems (Bryden and Caparini 2006), risks fragmentation, and diminishes authority (Krahmann 2003). Recent scholarship has highlighted this pattern in global financial governance as non-state actors gain infrastructural power (Mann 2008; Rodima-Taylor and Grimes 2019; Braun and Gabor 2020).

Other scholars argue that a proliferation of governance actors at different levels creates an assemblage that might provide greater coverage of an issue and thus enhanced governance (Daase and Friesendorf 2010). This transnational, multilevel network also creates the ability of powerful states to force compliance via “weaponized interdependence” (Farrell and Newman 2019). Others argue that IOs have a vital role to play in such settings as “orchestrators” (Abbott et al. 2021).

These broader debates align with debates attempting to explain the surprisingly high level of commitment to compliance with FATF standards. For some, the credible threat posed by the lists plays a central role in compliance (Daniel W. Drezner 2007; Eggenberger 2018; Morse 2022). Others point in particular to the US’s central informational role in the FATF network (Jakobi 2013). Still others show that the United States is able to force “cooperation” around illicit finance, even when FATF is unable to get that cooperation (Farrell and Newman 2019).

Other scholars explain AML cooperation in more managerialist terms. The regime’s very existence stems not from enforcement but from “ontological persuasion” (Hülsse 2007), or FATF’s successful efforts to persuade actors that money laundering is a problem worthy of attention and effort. FATF’s use of legal discourse constructs FATF as an authority (Abbott and Snidal 2000) and FATF’s ability to market practical knowledge as the “correct” solution (Hülsse and Kerwer 2007). To the degree that there is a cost for non-compliance, that cost extends first from the construction of those jurisdictions as bad actors (Sharman 2009).

The impact of multi-level, multi-actor governance assemblages is also significant. As noted above, FATF ultimately depends on banks to implement national interpretations of international standards, which in effect deputizes banks as front-line security agents of the state (Amicelle and Jacobsen 2016). For some scholars, this extended, indirect system of implementation severely limits FATF’s potential impact (Gutterman and Roberge 2019), while for others, this is how the lists get their “bite” (Sharman 2009). Still others see FATF as an orchestrator of the AML regime (Findley, Nielson, and Sharman 2015b).

FATF and the AML/CFT regime, then, reflect changes in global governance that scholars have noted in other issue areas. Like those other areas, scholars of the AML/CFT regime debate which of these dynamics are most important in shaping outcomes. Most take as their puzzle the spread of the AML/CFT regime noted in the introduction. While the frameworks all generate some useful insight, they seem insufficient to explain the contradictory contours of FATF’s impact that we show below: a belief in a systematic impact without much evidence for it. In this sense, as we discuss next, works that place ideas at the center of analysis provide an alternative explanation.

“Getting to Yes,” But Not Much Deeper: Compliance as Rational Myth

Work across disciplines (finance, sociology, and IR) and pitched at different levels of analysis (micro-, mezzo-, and macro-) focuses on the pivotal role of ideas in explaining global financial governance. On the micro-level, research on the decision-making of financial analysts emphasizes their limited rationality (Barberis and Thaler 2003; Cipriano and Gruca 2014; Duong, Pescetto, and Santamaria 2014). On an institutional level, ideational frameworks challenge broader assumptions about the rationality of institutional design and highlight how the effects of institutions often veer from the intentions of designers in ways that the rationalism of principal–agent modeling cannot fully explain (Wendt 2001; Weaver 2007). Others emphasize the potential within those institutions for real persuasion, defined as changing preferences regarding the ends, not just the means, which eliminates the need for enforcement (Checkel 2001; Gheciu 2005; Nance 2018).

Much work on the structural effects of ideas aims to explain patterns of continuity and change (Schmidt 2008; Campbell 2021). Building on the legacy of Ruggie (1982), scholars have shown, for example, how ideas underlie the deep shifts in what is considered “correct” economic policy (Blyth 2002; Linsi 2020), how international organizations promote diffusion via socialization (Abdelal 2007), and how international financial power is shaped by domestic legitimacy, not just the size of the market (Seabrooke 2006). Perhaps because they look in particular to crises as critical moments for understanding the role of ideas, much recent work on ideas within the political economy emphasizes how ideational dynamics can lead to perverse outcomes.

These multi-level ideational explanations fit well alongside a longer-standing intellectual project rooted in sociology on “rational myths” (Meyer and Rowan 1977; Boiral 2007; Colgan 2014). The framework challenges Coaseian and Weberian understandings of firms and bureaucratic organizations as rationalized equilibria that mitigate coordination problems (Shulock 1998). Instead, scholars argue that organizations unreflectingly integrate “the practices and procedures defined by prevailing rationalized concepts of organizational work and institutionalized in society” (Meyer and Rowan 1977, 340). Organizations are “manifestations of powerful institutional rules which function as highly rationalized myths” that become obligatory for organizations seeking to survive within a system (Meyer and Rowan 1977, 343). These myths are “rationalized and impersonal prescriptions” that portray “social purposes as technical ones” and lay out rules for rationally pursuing those supposedly technical goals. The myths are beyond the discretion of any one person or organization, and their legitimacy is largely taken for granted, such that close evaluations of the actual impact are more often assumed than evaluated (Meyer and Rowan 1977).

It is through conformance with the broader institutional rules, and not through a demonstrative of organizational effectiveness, that organizations increase their legitimacy and the prospect of organizational survival. As a result, various institutional components are “only loosely linked to each other and to activities, rules are often violated, decisions are often unimplemented or have uncertain consequences, technologies are of a problematic efficiency, and evaluation and inspection systems are subverted or rendered so vague as to provide little coordination” (Meyer and Rowan 1977, 343). In other words, institutions may adopt structures and systems that project “rationalism, formalism, and intellectual rigour,” but it does not mean that those systems attend directly to the problem the institution supposedly aims to address (Boiral 2007, 128). Compliance is more often ceremonial than substantive (Meyer and Rowan 1977). Sustaining this dynamic over time is a function of actors within the institution continuing to believe in the myth (Dick 2015).

Scholars of economic organizations have found the rational myth framework useful. For example, political leaders delegate monetary policy to independent central banks because of a general norm that legitimates it, rather than for the efficiency gains that economists see (McNamara 2002). Despite clear evidence that the Organization of the Petroleum Exporting Countries (OPEC) is unable to set global oil prices, its members often act as if they can, as do oil-importing countries, which grants members domestic and international power (Colgan 2014). Corporations symbolically conform to pressure to apply international environmental standards, but fail to ensure that daily practices change in significant ways (Boiral 2007; Boiral, Talbot, and Brotherton 2020).

In short, this scholarship suggests that we should not be surprised if form does not follow function, if organizational compliance is more ceremonial than substantive, and if actors within those organizations are motivated by external legitimation, but are blind to organizational impact. The remainder of this article shows how accurate these expectations are of blacklisting and compliance within the AML/CFT regime.

Research Design and Methods: Re-Examining the Impact of Blacklisting

Using a mixed-methods approach, we re-examine the case for blacklisting’s centrality in the AML/CFT regime. The FATF lists are a most likely case for materialist explanations of state behavior (Flyvbjerg 2006). States have honed the listing process over the years to enhance the lists’ impact. The regime has strong support from the United States, EU, and the United Kingdom, which together possess an overwhelming proportion of global financial power. Many of the targeted states are small or medium-sized economies that are heavily dependent on international banking to ensure access to FDI, foreign capital markets, and the flow of remittances. The process generates substantial information about states’ policies.

We begin by providing evidence that many observers expect the lists to have a negative financial impact. We re-examine the strongest evidence to date for a financial impact (Morse 2019, 2022) and find that those results are sensitive to the inclusion of more cases and the extension of the longitudinal analysis. We attempt to discover an effect of listing on other financial flows and again find no negative, statistically significant, systematic effects.

We then attempt to explain these null results. Using an abductive approach (Tavory and Timmermans 2014), we examine the decision-making and action of banks. We draw on existing scholarship and expert interviews.

This design avoids the challenges of endogeneity common in the study of threats, in which survivor bias often means the analyzed sample likely differs in some key way from the population (Smith 1995; Morgan, Bapat, and Kobayashi 2014). Our dependent variable, the financial impact of listing, involves the decisions of financial institutions, which as actors are separate from those who might drive anticipatory effects (i.e., policymakers).

Evidence

Expecting an Impact

The expectation that the FATF lists harm targeted jurisdictions is widespread.³ Scholars studying the subject commonly repeat the idea. For example, Findley, Nielson, and Sharman (2015b) argue that FATF’s reporting requirements and the blacklists give FATF “sharper teeth” than relevant UN conventions and that “FATF blacklisting induced tax havens to bring offending firms into line with global standards” (292). Drezner (2007) writes that, of FATF, the OECD, and the Financial Stability Forum, “the FATF initiative to enforce anti-money-laundering standards functions as an exemplar case of club standards enforcement.” Drezner continues: “There is clear evidence to support the contention that these jurisdictions altered their laws in direct response to the FATF threat of economic coercion.”

Policymakers also expect an impact, routinely expressing in interviews that fear of being listed is an important part of the regime (Nance 2015). Industry groups and policymakers in the Philippines and Thailand pushed for domestic reform out of fear of the lists (Morse 2022, 153–53, 157–59). Officials in Dominica and Russia cite avoidance of FATF blacklists in their push for economic reform (Drezner 2007). Sharman (2011, 99) recounts one official as saying that “the prospect of being blacklisted concentrated policymakers’ minds like ‘a gun to the head’.” The lists are increasingly visible in the financial press and ratings agencies sometimes take note.⁴ The support of the United States, the United Kingdom, and EU means that the vast preponderance of financial power lies with the regime, further lending credibility to this belief.

States’ actions confirm this expectation. States hire consultants to help them prepare for FATF reviews (Tsingou 2022). They reluctantly comply to pre-empt listing or to be de-listed (Vlcek 2010; Morse 2022). FATF and its members have consistently and conscientiously aimed to hone the blacklisting process in order to enhance the credibility of the threat the lists carry (Morse 2022). At least in reputation, this seems to be paying off. FATF engages states and non-state actors in various ways, but FATF is perhaps now best known for the lists. Google search analytics show that, for the time period we examine, three of the top six queries with “FATF” were about the lists (“list,” “fatf grey list,” and “grey list”).⁵

Reviewing Existing Evidence

Expecting a material impact from the lists is reasonable, as there are several ways that the lists might impose material costs.⁶ That includes the direct withdrawal of funds (Kudrle 2009; Morse 2019), slower or more expensive transactions (Sharman 2009), or the threat of foregone future transactions (Morse 2019). Compliance professionals play an expanding role within banks, too, which should mean that AML increasingly shapes business decisions (Tsingou 2018). The lists might also jeopardize much needed development aid (author interview with banking official 2020). These multiple pathways to harm arguably boost the credibility of threats.

Nevertheless, evidence for that impact is mixed. Small- and medium-N studies suggest as a whole that FATF’s lists have not been that effective. For example, Nauru reluctantly complied with FATF after being listed (Eggenberger 2018). But Nauru also faced a unilateral embargo by the United States, which most listed jurisdictions do not face. Work on earlier iterations of the blacklisting process shows that some jurisdictions complied, others only partially, and some ignored the lists (Drezner 2008). Blacklisting prior to 2010 had no apparent effect on the exchange rate between the US dollar and the currency of listed jurisdictions (Nance 2015). Members also quickly abandoned previous iterations of blacklisting in favor of new processes, suggesting a certain dissatisfaction (Nance 2015). Some find that a country’s preferences on more generalized norms, like multilateralism, explain more change than the lists (Pursiainen 2022). Targeted countries in sub-Saharan Africa interact strategically with FATF, projecting compliance while in fact doing little to fight illicit finance in their systems (Azinge-Egbiri 2021).

Statistical analyses are more mixed. Kudrle (2009) examines the effect of FATF lists on bank and non-bank assets and liabilities held vis-à-vis tax havens. He finds mixed results, including some cases of increased assets/liabilities after listing, and concludes that investors base their behavior on other factors. Others find increased flows after listing and suggest the lists have an advertising effect (Masciandaro 2005). Balakina, D'Andrea, and Masciandaro (2017) find no “stigma effect” on international capital movements among 126 countries, but the study includes different versions of the blacklist, making the findings difficult to interpret.

In contrast, Farías and Almeida (2014) find a reduction in the ratio of FDI to GDP among jurisdictions that are greylisted, although a sample of only seven greylisted countries raises concerns. The strongest support to date for the market enforcement hypothesis comes from (Morse 2019). Employing quarterly data from the Bank for International Settlements (BIS) to examine the effect of blacklisting on bank liabilities for 2010–2015, Morse (2019, 2022) finds that listing is negatively associated with bank liabilities.

Finally, very little research traces the causal pathway that we would expect blacklisting to follow, namely, whether financial institutions incorporate the lists into their decision-making. The few studies that do are not sanguine about the impact. Global field experiments on bank behavior show that banks will provide financial services to fictitious clients that are described in ways designed to trigger concern about money laundering and other illicit financial activity (Findley, Nielson, and Sharman 2014, 2015a). While there is variation, the authors interpret the findings as showing quite low levels of compliance within banks: “[T]he results of our experiments suggest material self-interest remains an all-too-powerful temptation to violate international standards” (Findley, Nielson, and Sharman 2015a, 160). Importantly, banks in the financial hegemons are among the worst offenders. Other research, to which we return below, shows that the informational foundations of bank decision-making are too diverse and too conflicting to expect any one list to drive larger patterns (Amicelle and Jacobsen 2016). These findings sit awkwardly alongside arguments suggesting that banks so closely adhere to FATF lists that they strongly shape international financial flows.

In sum, scholarship about the source of the AML/CFT regime’s influence remains inconclusive at best. In the following section, we contribute statistical and qualitative analyses to this debate.

Replication and Extension of Quantitative Data

(Re-)Examining the Effect of Blacklisting on Cross-Border Liabilities

We begin with the statistical evidence presented in Morse (2019). Morse’s model uses BIS data⁷ with country-fixed effects, a time polynomial to control for time trends, and lagged covariates. The results show a negative correlation between listing and cross-border banking liabilities, which supports the market enforcement hypothesis.⁸ Several characteristics of the original model compelled us to look more closely at the findings. First, it is the most direct claim in support of the market enforcement hypothesis. Second, the results suggest that listing results in a 14 percent decrease in baking liabilities, notably large considering the median decline in bank liabilities during the Great Recession (from 2007 to 2009) was 10.04 percent.⁹ Third, due to missing data in the covariates, simple listwise deletion drops the full sample of forty-seven countries down to only ten and the overall sample size from 3,288 to 656 country-quarter observations. This strategy is a common way to deal with data missingness, but it also risks biasing the results if the missingness is non-random. Fourth, the model relies on an estimation strategy akin to TWFE. Recently published research highlights the empirical problems with using TWFE in cases of time-varying treatment (Goodman-Bacon 2021). We therefore examine the sensitivity of the analysis to an alternate modeling approach that addresses the identified problems with using TWFE with time-varying treatment (Callaway and Sant'Anna 2021). Fifth, as seen in Online Appendix Figure A.1, the level of log liabilities of listed and non-listed country quarters, on average in the full sample, does not differ. There is also an over-representation of lower values of logged bank liabilities in the original model sample compared with the distribution of the full sample, which could contribute to a downward bias. Finally, so many observations are missing for some countries that there are no quarters where they are not being listed, and so the level of bank liabilities pre-/post-listing cannot be compared in the model.¹⁰ Several listed countries also enter the period of analysis (2010) already on the list, and so the pre-/post-listing comparison is not available.

Table 2 Model 1 presents the replicated results of the original research (with robust standard errors).¹¹ To test for sample bias induced by data missingness, in Model 2, we remove the controls of Model 1, running a bivariate regression of just Listing on bank liabilities among the same sample used in Model 1. The effect of Listing in Model 2 remains negative and significant and of a similar magnitude as in the full model (Model 1). This suggests that the control variables are not controlling for much additional variation. We can therefore generate a much larger sample size through bivariate regression.¹²

Model 3 shows the results of the same bivariate regression as Model 2 but using the full sample of countries that the bivariate analysis allows by avoiding the missingness in the controls. This fuller sample includes all forty-seven countries listed at any point in the period of analysis and the eighty-eight never-listed countries, which increases the observations from 656 in Models 1 and 2 to 2,680. In Model 3, the effect of Listing is no longer statistically significant. Additionally, the smaller effect size suggests a “negligible effect” (Rainey 2014) of only a 2.3 percent decrease.¹³

Finally, we then use multiple imputation to fill in missing observations in the covariates in Model 4. Although recent studies demonstrate the limited utility of multiple imputation as a solution for missing data (Arel-Bundock and Pelc 2018; Pepinsky 2018), we add this model as a robustness check following standard practice for handling missing data (Lall 2016). To partially account for potential non-random, systematic missingness, we remove eleven countries with all or most observations missing for more than three covariates.¹⁴ Still, the effect of Listing in this model is insignificant. Thus, with and without multiple imputation, the results suggest that blacklisting does not lead to a widespread, systematic outflow of funds.

Looking Beyond Cross-Border Liabilities

Of course, it is possible that blacklisting’s financial harm manifests elsewhere. To test for this possibility, we use a similar modeling approach to look for the effect of listing on other financial flows. We use the International Monetary Fund’s cross-border portfolio investment flows data to examine the effect of listing on portfolio inflows and outflows. Portfolio investments may be considered a more dynamic class of assets given that it captures all tradable securities that have relatively short-term lengths and therefore may better reflect the dynamic responses of investor perceptions of overseas markets.¹⁵ We also test the effect of listing on FDI and World Bank development aid to see if material financial impact is occurring through other financial channels.

Table 3 presents the results of the IMF portfolio inflows and outflows data models, where we use the same modeling specification as the original model (Models 1 and 2), bivariate results (Models 3 and 4), and multiple imputation results (Models 5 and 6). In each model predicting portfolio inflows, the effect of listing is insignificant. Meanwhile, in Models 4 and 6 predicting portfolio outflows, the effect of listing is significant and negative—suggesting capital is more likely to stay than leave after listing. Collectively, the results suggest that blacklisting does not generate a systematic, negative impact on these major categories of cross-border financial flows.

Finally, we test the effect of listing in FDI and World Bank aid and development assistance both as a bivariate regression and a full model using multiple imputation as above. The results are presented in Online Appendix Table A.4 (p. 4). We do not observe a significant effect of listing in FDI in either the bivariate or multiple imputation context. We do observe a significant negative effective listing on World Bank aid and development assistance in the model using multiple imputation, in line with de Koker, Howell, and Morris (2023). However, this effect disappears without multiple imputation, reflecting the limits of multiple imputation cited above. More importantly, as the next section will show, this effect is not significant when using a more rigorous modeling approach that addresses fundamental issues with the use of TWFE in time-varying treatment (see Table 4).

Sensitivity to New TWFE Modeling Approach

Finally, we examine the sensitivity of these results to using a modeling approach akin to TWFE and extend the sample of listed countries based on our own coding of FATF reports from 2010 through 2020. Recent research identifies significant issues with TWFE models with time-varying treatment—i.e., in this case, using country-fixed effects and a time polynomial (which makes it akin to TWFE) with multiple varying periods of Listing. As described in Goodman-Bacon (2021), varying treatment creates multiple pre- and post-treatment time periods and multiple treatment “groups.” The upshot of these varied comparison groups is that the TWFE difference-in-difference estimator is difficult to interpret, can produce biased estimates, and can only be interpreted causally under strong assumptions about the treatment effects (Goodman-Bacon 2021). A new approach addresses these issues by effectively subsetting the data, estimating the average treatment effect for each treatment group (basically reverting the estimation to the basic difference-in-difference setup), and recomposing these group effects into aggregated effect estimates (Callaway and Sant'Anna 2021).

In order to use this new method, we add a 10-year pre-treatment period (beginning in 2000) using the same set of outcome variables as above and collecting control data that captures similar measures as in Morse but with minimal missingness. Despite extending the pre-treatment period, we conservatively exclude countries involved in listing periods prior to 2010 to avoid possible lingering reputational costs in the later listing period, including the “true” blacklisted countries Iran and North Korea.¹⁶ We further note two issues that affect interpretation of the model results. First, the Callaway and Sant'Anna approach does not allow for treatment to turn on and off; once a unit is treated, it is treated forever. Yet several countries go on and off the blacklists, with some countries going on a second time after being removed for a period. We therefore limit our interpretation of the dynamic results to less than 4 years after treatment (the average length of listing).¹⁷ We also note that in this expanded time range of listing (2010 through 2020), most treatment “groups” are quite small, especially at the quarter level, as shown in Figure 1. The group average treatment effect—the building block of the Callaway and Sant'Anna approach—is therefore difficult to recover reliably for these groups. We present the results with quarterly and yearly data. In both cases, we further limit our interpretation of the results based on this sample size consideration.

In presenting our results, we focus on two aggregate effects: the overall average treatment effect and the aggregate treatment effect over time (event study or “dynamic” treatment effects). The Callaway and Sant'Anna approach also allows for the inclusion of time-varying covariates, which are otherwise another complicating factor in the two-way fixed effects difference-in-differences (TWFEDD) estimation. Time-varying covariates are only incorporated in the analysis in the pre-treatment period, where their importance is primarily in establishing conditional parallel pre-trends if the unconditional parallel trends assumption is not met.¹⁸ The necessity of including covariates to establish conditional parallel pre-trends can be observed using the pre-treatment dynamic effect estimates. If they are not needed—if the pre-treatment dynamic effect estimates are insignificant—the dynamic effects can be interpreted as valid causal parameters capturing the aggregated difference-in-differences estimates of the different treatment groups over time.

Figure 2 presents the dynamic results of the model when run with the full set of quarterly and yearly observations. Statistical significance is indicated by the “whisker” falling entirely either above (for positive relationships) or below (for negative relationships) the line at 0. The fact that the pre-treatment effect estimates (to the left of time 0) are all insignificant suggests we do not need to control for conditional parallel trends (and therefore do not need to include covariates). Each estimate to the right of time 0 (the onset of treatment) corresponds to the effect of listing on bank liabilities in each period since treatment.¹⁹ This allows an interpretation of the effect of treatment over time. Under a strict interpretation of the market hypothesis, wherein banks are responding dynamically to a country being listed, we would expect to see a relatively quick (less than 1–2 years) negative and significant effect of listing on bank liabilities. We do not observe significant negative effects even several years following listing in either the quarterly or yearly data.

As a robustness check to the models in Figure 2, we rerun the models with several of the same covariates as in the replication model above, despite the fact that this significantly drops the sample size.²⁰ We do not find significant effects. We also run the models using the “not yet treated” groups rather than “never treated” as the comparison group, which also do not yield significant effects.²¹

Finally, we run these same dynamic models with different operationalizations of cross-border investments as the outcome variable as above. For the sake of space, these results are presented in the Online Appendix.²² In none of the specifications do we find evidence for dynamic effects of listing on either portfolio investment flows or World Bank aid.²³ Supporting the methodological argument of this paper, the latter of those was significant using a less rigorous modeling strategy. The total average treatment effects for all three outcome variables are included in Table 4, none of which are significant.²⁴

While technically the estimates in Table 4 could be interpreted causally—or rather, we could claim to not find a causal effect—we are more reserved in our interpretation of the results given the barriers related to the limited size of the treatment groups. More importantly, we do not feel claiming a causal argument is necessary to emphasize our broader point: The claim of a systematic impact of blacklisting on financial outcomes is difficult to justify.

In short, despite looking for evidence across different measures of financial flows, running the model with and without imputed data, and adjusting the TWFE strategy to reflect recent methodological innovations, we find no evidence of a systematic impact from listing. The remainder of the paper turns to our further qualitative investigation of what might then explain this lack of observable effect.

Tracing the Path of Impact

Qualitative Approach

Recall that the AML/CFT regime “deputizes” banks and other financial institutions, requiring them to serve as front-line agents responsible for applying national interpretations of international standards. Thus, the FATF lists (and others like it) can be understood as messages passed via a signal that the intended audience has to receive, interpret, and respond to (Amicelle and Jacobsen 2016). The expectation that listing causes financial harm requires that FATF persuasively communicate that listed jurisdictions are non-compliant and high-risk and that financial institutions respond by limiting financial relationships with entities in those jurisdictions. The null results above suggest that this is not happening. To understand why, we focused on the decision-making of those front-line agents, the banks.

Evidence on this question is difficult to attain. Banks are not transparent about how they handle potentially illicit activity. Secondary research on the topic is also limited, although some exists. In combination with that secondary research, we use expert interviews to generate causal process observations, or “crucial insights into the causal processes of interest” (Freedman 2010, 190).²⁵ Because our aim is process tracing, a representative sampling strategy is not necessary, nor do we need dozens of interviews (Lynch 2013). Rather, the explanatory power of the interviews stems from their role in our triangulation strategy and from the expertise of our informants. We used a snowball sampling strategy designed to include experts who would be able to speak to trends in the industry. In semi-structured interviews, we probed how banks make investment and client onboarding decisions in the context of KYC obligations and whether FATF figured into the calculus.

We began with a well-known AML expert whose career has included working on nearly all sides of AML: with the government, as both a law enforcement official and as a regulator; in the private sector, including high-level stints at some of the world’s largest financial institutions; and with (although not for) FATF directly. That informant provided us contacts for interviews with representatives from companies that help banks conduct customer due diligence (CDD). These interviews included representatives from one of the world’s most centrally situated financial services organizations. Some of the informants began their AML careers in government and then transitioned to private sector jobs; others have always been in the private sector. We conducted and recorded these interviews (n = 6) via Zoom. They each lasted roughly 90 min. Informants also led us to additional documentary evidence. In two interviews, informants gave us a demonstration of the technology that they use to conduct research on banks’ clients. Both authors participated in each interview, and we followed each interview with a de-briefing that we recorded.²⁶ One author conducted one additional interview in person.

Tracing Decision-Making in Financial Institutions

The interviews and secondary sources generate three key insights that explain why the FATF signal does not generate the expected financial impact and thus give us greater confidence in the null results above. In brief, the signal is one among many, is noisy, and is faint.

First, FATF’s signal is one among many, generally conflicting signals about customer risk that financial institutions must receive and interpret. One seemingly frustrated informant declared: “the OFAC list, the EU list, the Bank of England’s got a list, the UN’s got a list, I mean, you could go on and on.”²⁷ This aligns with work by Amicelle and Jacobsen (2016), who quote another senior compliance officer as commenting: “Screening lists? With our own intelligence systems, at the moment we run 35 pages of names of lists, twelve lists on a page, and that's just the titles of the lists” (89). One company we interviewed advertises in particular its ability to assess “country risk,” which is the focus of FATF’s lists. FATF is part of the company’s assessment. But so are country risk ratings from Transparency International, the World Bank, the US State Department, and the United Nations.

Vitally, the lists differ substantially from each other. There is some overlap across, for example, the EU AML blacklist and FATF’s. But the lack of consistency across lists is such that some scholars argue the inconsistency renders the regime ineffective (Riccardi 2022). Those lists also differ in their scale and scope. Some focus on sanctions, good governance, money laundering, or human rights. Some name just the worst performing, while others rank all jurisdictions. Even the lists focused on money laundering reveal an underlying lack of agreement on what constitutes money laundering risk (Riccardi 2022). In that context—one of multiple lists, with varying targets and differing scales—it would be surprising if any one list, including FATF’s, so drove financial decisions that they were predictive of financial flows. One possible exception might be US and/or EU sanctions lists. Yet those are much more limited, and generally now more targeted, than FATF’s lists of countries.

A second insight generated by interviews is that FATF’s signal is relatively noisy when compared to other information sources. Our interviews revealed the sophistication of information to which banks, especially major international banks that drive financial flows, have access. Informants emphasized the role of third-party companies that specialize in helping other institutions meet their obligations to conduct thorough “KYC/CDD” investigations. Multiple informants, existing research, and sector growth confirm this is a general tendency in banking. Many of the names in this space are well known: McKinsey, EY, and Dow Jones. Others are smaller and specialize in only KYC processing. Some companies specialize in the platform that sorts the data, some specialize in just the data, and others offer both services and tailor it to clients’ preferences. The newest versions of KYC utilities have substantial artificial intelligence or machine learning components to them.

In two cases, informants demonstrated their software during our interviews. One informant showed us the web of people and companies generated from analyzing Gazprom, the Russian oil company. In the second, we observed as the informant tracked the business relationships that linked the exiled and sanctioned ex-dictator of Gambia, Yahya Jammeh, to a specific house in Potomac, MD, USA.²⁸ Reflecting the previous point, these companies do not only—or even primarily—aim to identify money laundering. They attempt to help financial institutions understand the level and kind of risk associated with an individual account. In many cases, that pertains to targeted sanctions.

These companies are private sector responses to the demand created by AML regulation. To return to an earlier metaphor, they are like private security guards hired by deputized banks to help them police their clients. Unlike banks, which are part of a formal delegation of state power and are overseen by financial regulators, KYC companies are not formally part of the AML regime governance chain.²⁹ As one informant emphasized, these KYC companies for the moment are completely unregulated actors in this AML space.³⁰

KYC companies provide banks that use their services with considerable amounts of client-level data. Some highlight that their list of politically exposed persons—just one small category of red flags—runs to 1.3 million individuals; some claim to include over 400 different lists (Amicelle and Jacobsen 2016). LexisNexis markets its widely used World Compliance database as including “over 5 million profiles of individuals and companies that are updated on a daily basis” with sixty risk categories, “global adverse media profiles from over 30,000 worldwide feeds,” and “over 450 researchers with fluency in more than 50 languages.”³¹ To this, banks add their own filters and data, especially regarding risk appetite.³² In a high-level policy forum discussing KYC in which one of the authors participated, the head of compliance at a major global bank boasted of the bank’s 43 petabytes (43 million gigabytes) of data on their customers.³³ The bank filters this client data through the broader business model of the institution, including their general risk profile. In that setting—where machine learning and specialized third-party companies make sense of 43 petabytes of data—FATF’s country-level list by comparison is a noisy signal.

A third insight from the interviews was that FATF’s signal is relatively faint. That is, among the many signals banks receive, FATF’s lists do not necessarily stand out. For example, KYC employees confirmed that FATF’s lists are not especially heavily weighted in their algorithms. The guidance on CDD from the Wolfsberg Group reveals a similar trend. The Wolfsberg Group describes itself as “an association of thirteen global banks which aims to develop frameworks and guidance for the management of financial crime risks.” To streamline KYC and information sharing, the Wolfsberg Group promotes standardization of information around their “Correspondent Banking Due Diligence Questionnaire” (CBDDQ). The CBDDQ is a series of 107 questions, some with sub- or even sub-sub-questions that are designed to help banks assess the risk posed by a client. FATF is referenced in nine questions, but the CBDDQ never references the FATF lists. One question asks whether a specific FATF recommendation regarding wire transfers is in place.³⁴ Others recommend that financial institutions should follow FATF’s guidance on, inter alia, beneficial ownership transparency, record keeping, or virtual currencies. In other words, the world’s most influential group of global banks, all of which are headquartered in states that are members of FATF, does not ask banks when making client decisions to consider whether a transaction might involve a listed jurisdiction. Finally, showing the reinforcing nature of this ecosystem of organizations, in an interview with one of the world’s most important financial intermediary organizations, the informants discussed their efforts to help different global regions develop common KYC standards. When asked whether FATF was an important part of the standards they were promoting, they said they relied on the Wolfsberg Group’s CBDDQ.³⁵

To all of this, we can add the exemplary and startling research of Findley, Nielsen, and Sharman (2014, 2015a), referenced above. In global field experiments, the researchers tested whether banks would provide financial services in response to requests such a way as to elicit suspicion based on FATF lists, including listing FATF by name. They find that naming FATF while also asking for services decide to contradict FATF’s rules “had no significant effects on response or compliance rates” (2014, 2015a, 148).

In sum, evidence gathered through secondary research and interview-based process tracing confirms the null results we report above. This closer look at bank decision-making flips the puzzle, making it seem highly improbable that FATF’s lists would generate the kind of systematic impact that so many observers expect.

Analysis and Conclusion: The Political Impact of Rational Myths

The statistical and qualitative evidence above provides strong support for the argument that blacklisting in FATF does not have the widespread, systematic financial impact that many expect. Once sample and modeling issues are resolved, correlations previously found to be statistically significant no longer are. Nor do additional measures of financial flows show significant associations with FATF’s lists. Process tracing through interviews and secondary sources about banks’ decision-making helps interpret the lack of evidence for an impact. When making decisions about clients, banks consider tens or even hundreds of different lists. Many of them have access to massive amounts of client-level data, making the country-level risk that FATF highlights less useful. As a result, FATF lists are not heavily weighted in decision-making. All considered, it would be surprising if the FATF lists had a systematic impact on financial flows. Yet many scholars, policymakers, and bankers expect just that.

The “rational myths” framework helps explain why institutional structures persist despite institutional performance that diverges from the logic that drove their creation in the first place. In a broader institutional context that emphasizes rationality in governance, organizations implement processes, structures, and standards that are designed to appear technical, rule-like, and impersonal. They and other organizations “ceremoniously reproduce” those standards (Boiral 2007, 128). Ceremonial adoption, however, often fails to generate meaningful change in actions (Meyer and Rowan 1977).

The FATF blacklisting process reflects this tendency. Members have “rationalized” or “legalized” blacklisting (Abbott et al. 2000); proponents argue that it is more technical, more automatic, and more transparent. There is a nearly 200-page Common Methodology. Mutual Evaluations are carried out by trained experts seconded from other international organizations. The supposed enforcement via third parties—“the market”—lends the process credibility. Yet at each level, actors proceed ceremoniously. FATF constructs the lists, states comply (Morse 2022; Tsingou 2022), and banks adopt FATF guidance on AML risk assessment (Amicelle and Jacobsen 2016; Tsingou 2022). All of this activity echoes the reasonably expectation that being listed harms the target’s economy impact, despite a lack of supportive underlying empirical evidence. The lists’ power, therefore, stems from a belief in their impact, which generates ceremonial compliance.

There is precedent for this finding in other issue areas. Colgan (2014) shows that states—members and non-members alike—respond to OPEC as if it were a true cartel. Similarly, the FATF lists generate efforts to comply and avoid blatant non-compliance. Officials respond as if the lists had widespread financial impact, which in turn likely further reinforces the belief for observers. In this sense, the lists discipline, at least partially, but rarely punish.

These findings raise a series of avenues for compelling future research on the AML/CFT regime and beyond. The lack of a systematic effect leaves open the possibility for other avenues of change or even enforcement. Perhaps some subset of countries are affected. If so, understanding when the lists cause harm might shed light on when they do not. Alternatively, that research might reveal instead that observers attribute economic downturns to FATF, when other dynamics are driving negative outcomes. As implied by the weaponized interdependence framework (Farrell and Newman 2019), it is possible that compliance stems from a fear of punishment by financial hegemons, not “the market” more broadly. Such findings, in combination with the null results reported above, would raise other questions about the selective enforcement of standards and whether and why that apparently higher level of scrutiny goes beyond enforcing the ceremonial compliance highlighted here.³⁶ Future research also could unpack how the construction of the lists interacts with other efforts within FATF aimed more at soft forms of power, including knowledge creation, learning, and persuasion. We remain open to the idea that change within the AML regime happens via the many other facets of FATF’s work, not via blacklist-based material enforcement.

Very intriguing, too, is the positive correlation found between listing and cross-border banking liabilities (Table 2, Model 4) and the negative correlation between outflows and listing (Table 3, Model 4). Others have found evidence of this “advertising effect” (Masciandaro 2005; Kudrle 2009), indicating that financial institutions and their clients might be interested in jurisdictions with lax financial regulation. Most research on illicit finance arguably has an unstated assumption that banks prefer to avoid laundering money. Our results indicate that this may not be accurate. More work also is needed to understand decision-making within banks and how that affects global governance efforts.

Beyond AML, the “most likely case” design of this project suggests that future research should more seriously examine institutions of global governance through the lens of “rational myths,” including global financial and economic governance. Under what conditions do myths develop, why, and to what effect?

The results also yield avenues for future work on “weaponized interdependence” by suggesting important limitations (Farrell and Newman 2019). The power imbalances entailed in FATF should maximize the potential for supporters to “choke off economic and information flows, discover and exploit vulnerabilities, compel policy change, and deter unwanted actions” (45). And while the lists may shape state compliance, they are less effective in shaping bank actions, which ultimately drive flows. The same is true of arguments that explain FATF as a case of orchestration (Findley, Nielson, and Sharman 2015b). It may be that the AML/CFT assemblage is too dispersed for orchestration to work (Krahmann 2003): The key players are too far from the conductor.

More fundamentally, these results suggest that scholarship continues to overlook the non-material underpinnings of the global political economy. The FATF lists examined here are explicit attempts to leverage the massive imbalance in market power held by FATF supporters. Compliance stemming from fear of the lists, based on a belief in financial costs not systematically borne out in the data, cannot be explained via purely material frameworks. If ideational factors play such a pivotal role in how this case plays out, then it seems likely that they are hard at work in a much wider swath of cases than IPE scholars have acknowledged to date.

Footnotes

Author Biography

Devin Case-Ruchala is a Postdoctoral Fellow in the Department of Political Science at the University of North Carolina at Asheville. They study the political economy of finance from an international, comparative, and historical perspective. Broadly, they are interested in the political constitution and governance of financial markets and employ both quantitative and qualitative methodological approaches. Their core research agenda explores the political economic origins and diffusion of government-initiated public banks (“G-Pubs”). In additional research projects, they explore the role of regime types in financial integration and financial crisis diffusion as well as global governance of illicit finance. They received their B.S. in Business Economics and Public Policy from Indiana University.

Mark Nance is an Associate Professor of Political Science in NC State University’s School of Public & International Affairs. He studies the politics of global financial relations and global governance in transnational illicit markets, with a particular focus on efforts to counter illicit finance and is co-founder of the Money Laundering Research Network.

Notes

Authors’ note: In addition to the thorough anonymous reviewers, the authors thank Cameron Ballard-Rosa, Navin Bapat, Mark Blyth, Matt Collin, Mark Copelovitch, Louis DeKoker, Daniel Drezner, Mike Findley, Boram Lee, Charles Littrell, Daniel Nielson, Jason Sharman, Eleni Tsingou, and Jim Zink. Julia Morse deserves a special word of thanks for her openness to debate. For additional insight, we thank participants in the UNC-IR Brown Bag Workshop, the UNC Methods and Design Workshop, NC State’s Political Science Writers’ Assistance Group, the Triangle Institute for Security Studies Paper Workshop, attendees of the Central Bank of the Bahamas 2nd Empirical AML Conference, the Midwest Political Science Association 2021, and ISA 2021 for their insights. We also thank Ilkka Malmi for invaluable research assistance.

References