Skip to main content

Money laundering and AML regulatory and judicial system regimes: investigation of FinCEN files


Using a novel dataset, this paper explores the link between cross-border flows of illicit money and anti-money laundering (AML) regulatory and judicial system regimes. To this extent, we explore the information contained in thousands of suspicious activity reports filed by US banks and recently disseminated within the FinCEN files investigation. For a sample of 106 jurisdictions, we relate money laundering (ML) flows as well as the central role of each country’s banks within the international ML network to several variables capturing the AML regulatory stance and ML enforcement. The results point to the crucial role played by judicial system performance variables in explaining the layering phase patterns observed in the underlying data.

This is a preview of subscription content, access via your institution.


  1. There are several stages of money laundering: placement, layering, and integration. The dirty money, once collected by criminals, is brought to a financial institution in the placement or prewashing stage. In this phase, several possible techniques are used by criminals to evade the AML regulatory radar, including breaking up deposits to values under the regulatory threshold, smuggling illicit money abroad, possibly after having exchanged the ill-gotten cash with lighter travelers’ cheques, and gambling at the casino to justify the provenience of the cash. After layering, the latter phase (integration) involves the investment of the laundered proceedings in real estate, financial assets, or luxury goods (see Teichmann, 2017, for a discussion and references therein).

  2. Network theory is seldom used in ML-related studies. An exception is Colladon and Remondi (2017) who apply social network techniques to money laundering using transaction data of an Italian factoring company. The authors show that network metrics can capture illicit transactions and, thus, help identify high-risk customers.

  3. The list of country reports is available from Financial Action Task Force (FATF) website, accessible from

  4. According to the US Department of State’s International Narcotics Control Strategy Report (INCSR), the key AML laws and regulations include banks’ requirements to: (1) do due diligence to uncover customers’ identities; (2) report suspicious transactions; (3) report large transactions; (4) record keeping of large and unusual transactions; and (5) record and report transactions related to the financing of terrorist activities.

  5. A comparative table for 2013 is found here:

  6. See Barone and Masciandaro (2008) for details.

  7. Ferwerda et al. (2020) also estimate, through simulation techniques, international money laundering flows. In this paper, however, we only relate to the first part of their results, that is, the identification of the country characteristics explaining money laundering flows.

  8. See Unger and Hertog (2012) for a comprehensive discussion of global anti-money laundering regulation frameworks.

  9. With regards to the crime rate, Ferwerda (2009) finds that the extent and toughness of AML regulations significantly curb the crime rate and illegal behavior.

  10. Relatedly, Masciandaro and Portolano (2003) theoretically show that blacklisting regulatory AML may have an adverse effect on bank efficiency because it may hamper the supply of superior-quality financial products.

  11. Some other papers look at ML and more general legal standards. Gnutzmann et al. (2010) theoretically show that low financial standards are more likely to be associated with countries tolerating money laundering, particularly if these are low income. Vaithilingam and Nair (2007) find that money laundering is less pervasive in countries with more sound legal and higher innovative capacity.

  12. However, since the 1996 “Know Your Customer” AML standard, US banks have been encouraged to report continuous patterns rather than isolated suspicious activities to skim the transactions that are most likely to be related to ML before sending reports to the FinCEN. By using standardized report forms, banks are also required to detail what makes the transaction unusual.

  13. Other measures of centrality, such as the degree of centrality and betweenness, are measures centered exclusively on the number of ties or links for each node. See the pioneering work by Freeman (1979) for an overview of the centrality measures in network analysis.

  14. The path length is a function of the ties. The higher the number of paths or sequence of ties from one country to another, the longer the path (direct links between nodes have a length equal to one).

  15. Whenever possible, we extract relevant judicial data for 2012 because this year precedes the highest number of observations in the FinCEN files (2013).

  16. AML regulation is measured as an index equal to 100 if all the AML regulations (see details in the Methodology Section) specified by the US International Narcotics Control Strategy Report (INCSR) apply.

  17. All the countries in our sample punish money-laundering drug trafficking and other illicit activities beyond drug trafficking (for instance, fraud and human trafficking).

  18. We use the 2013 report as most of our data refers to this year.

  19. Summaries of country reports are available here:

  20. The dummy is constructed using data available from the FATF website, section “Jurisdictions under increased monitoring”, using the earliest available list (2020). This list of jurisdictions that are cooperating but still depict deficiencies in AML regimes is also called the “gray list”. The FATF’s earlier efforts to identify high-risk countries involved a more severe blacklisting approach and date the back to early 2000s. This list has been highly challenged particularly because of its erratic and opaque nature (see Hülsse, 2008, Ferweda et al., 2019).

  21. We consider destination flows or ML inflows. Within the FinCEN files, ML inflows and outflows are highly correlated on a country-by-country basis, preventing the carrying out of a disentangled analysis of in-and-out ML flows. Also, we are unable to say with certainty whether outward flows are towards a final destination or are just passing through.

  22. See Argentiero et al. (2008) and Masciandaro (2005) for a discussion of how ML is affected by economic growth.

  23. A closer look at the data reveals that a positive relationship between AML_def and ML is not necessarily at odds with the weakly positive coefficient of the FATF recommendation outcomes. There are a number of jurisdictions in our sample that have a very low FATF_recom score and are not listed by the FATF as high-risk jurisdictions. This is the case for several African countries, such as Congo, Sierra Leone, and Ivory Coast, complying with only a handful of FATF recommendations and not grey-listed. Ferweda et al. (2019) also highlight this ambiguity of the FATF strategy and provide a discussion on the criteria the FATF uses to identify high-risk jurisdictions.

  24. We use the variable cross-border debt collected by the Bank for International Settlements in 2013 year-end (locational banking statistics).

  25. Peripheral countries are here conceived as those jurisdictions that play a more passive role within the network as they do not send any ML flows and receive the latter from only a handful of other countries. Overall, we identify 12 countries in this group. These are Ecuador, Guernsey, Jersey, Andorra, Belarus, Croatia, Jamaica, Madagascar, Pakistan and British Virgin Island, Monaco, and Seychelles.

  26. This is a general specification for ML centrality using three-year average values for all variables and prosecution rate as a proxy for efficiency of the ML law enforcement (details about the construction of the variable at the bottom of Table A.5). These estimates, although overall confirming our baseline results, must be treated carefully as the partial nature of our data may lead to biased and inconsistent estimated parameters.

  27. We find that in the specifications in which pros enter, the estimated coefficient of the latter variable is negative and significant. The two indexes capturing AML regulatory stance are overall not significant across specifications (except for a marginally significant positive coefficient in column 2). In column (2), when considering ML_flows as the dependent variable, the cross-border flows regressor is still not significant, confirming the results advanced in Table 5 for the centrality regression.


Download references


The authors did not receive support from any organization for the submitted work.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Carmela D’Avino.

Ethics declarations

Conflict of interest

The author declare that there are no conflicts of interest related to the writing of the paper.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.



See Tables

Table 7 Literature review summary


Table 8 Overall value of SAR transactions in the FinCEN files, aggregated over all countries, by year, in USD


Table 9 Variables sources and details


Table 10 Descriptive statistics

10 and

Table 11 Additional robusteness specifications


Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

D’Avino, C. Money laundering and AML regulatory and judicial system regimes: investigation of FinCEN files. Eur J Law Econ 55, 195–223 (2023).

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


  • Money laundering
  • Anti-money laundering regulation
  • Cross-border money laundering flows