Abstract

This paper investigates the patterns of business ownership in Europe, using a unique dataset on the nationality of 28.7 million shareholders of companies registered in 41 European countries. By means of an exploratory multivariate analysis, it tests whether ownership links between different countries are driven exclusively by social and macroeconomic variables—such as trade or geographic or cultural proximity—or instead are also related to measures of financial secrecy, corruption and lack of compliance to anti-money laundering regulations. The results indicate that factors other than licit economic incentives explain the international ownership structure of European companies. European firms have an abnormal number (i.e. above the predicted value) of owners from tax havens and countries with poor financial transparency, which may suggest the use of holding companies for money laundering, tax evasion and to conceal illicit financial flows. However, ceteris paribus, the number of owners is abnormal in countries where rule of law and the control of corruption are more effective, suggesting that high level of corruption may be a cost in money laundering activities. The findings contribute to the current international debate on illicit financial flows—as framed by United Nations SDG 16.4—and can be used by public agencies and private actors to detect anomalies in business ownership and prevent potential financial crime schemes at corporate level.

Introduction

In recent years, a global consensus has emerged on the need to increase the transparency of company ownership to prevent transnational crimes (European Commission 2017; FATF 2016). Criminals can hide their identities or their illicit proceeds behind a veil of complex and extensive corporate structures, often set up offshore. As stressed by recent media investigations such as Panama or Paradise Papers (ICIJ 2017, 2018), opaque corporate entities can be used for tax evasion and tax avoidance (Alstadsæter et al. 2017; Cobham and Janský 2017; Zucman 2013), to conceal large-scale corruption schemes (van der Does de Willebois et al. 2011), to launder money (Savona and Riccardi 2017; Unger et al. 2014), and to facilitate transnational organised crime (Savona and Riccardi 2018; Steinko 2012). Therefore, more knowledge on (especially cross-border) business ownership is of utmost importance in detecting and preventing illicit financial flows (IFFs).

This need has also been stressed by the FATF (2012) Recommendations and acknowledged at European Union (EU) level in the updated versions of the anti-money laundering (AML) Directive (EU Directive 2018/843). The EU AML regime requires (i) obliged entities—such as banks, notaries and other professionals—to investigate the ownership structure and identify the beneficial owners of their customers; and (ii) EU Member States to set up central public registers of beneficial owners (Art. 13:1(b)). Despite the regulatory developments, knowledge about who owns European businesses remains scant. Just as scant is the understanding of the extent to which business ownership connections with certain foreign countries are ‘risky’.

This paper addresses this gap in knowledge by analysing cross-border ownership links as ‘red flags’ of IFFs. This study is one of the first empirical analyses of the determinants of companies’ cross-border ownership links, and one of the first empirical investigations of the role played by financial secrecy and effective governance in shaping the transnational structures of IFFs. We define cross-border ownership link or cross-border shareholding as any case in which a shareholder—either a natural or legal person—from a j-country holds a share in the share capital of a legal person registered in an i-country, when i is not equal to j (see Section 1.1 for an example and details). The focus of this paper is on legal persons only and the unit of analysis is the country, i.e. the aggregate volume of ownership links between legal persons located in i and j. In accordance with most of the literature (e.g. Aziani 2018; UNODC 2017), with the expression illicit financial flows, we refer instead to the transnational movements of assets related to illicit activities such as money laundering, corruption (especially grand corruption) and tax evasion schemes—eventually, including some forms of tax avoidance.1

Using an exploratory multivariate analysis of data on the nationality of shareholders of companies registered in 41 European countries, the paper analyses whether cross-border business-ownership links are exclusively driven by legitimate determinants—e.g. economic size, cultural proximity, trade or other social and macro-economic drivers—or; whether they are explained also by illicit determinants, related to countries’ financial secrecy and level of corruption.

The paper tests two hypotheses: (H1) that, after controlling for legitimate determinants, cross-border ownership links are explained by the financial and corporate secrecy of the jurisdiction in which the shareholder is located: the higher the level of secrecy, the more likely a legal person will be registered in that country and control a firm in another (more transparent) country; (H2) that, ceteris paribus, cross-border ownership is negatively correlated with the level of corruption of the country in which the shareholder is located: shareholders, for setting up holding companies, will look at countries with higher secrecy but lower corruption. In other words, we hypothesise secrecy and corruption to be inversely correlated, meaning that high corruption level represents a cost for people willing to design an IFF scheme.

To test these hypotheses, the paper develops a methodology, based on a gravity model, which makes it possible to detect ‘anomalous’ cross-border ownership links, i.e. ties which are abnormally above the value predicted by legitimate determinants such as geographic, social and economic relations between two countries, and which therefore suggest that corporate entities are possibly used to manage IFFs. The ranking of anomalous links produced by this paper can help countries to identify more precisely those foreign jurisdictions on which to focus their monitoring and investigation resources and with which to strengthen international police and judicial cooperation.

The paper is structured as follows. Section 1 provides a literature review and presents our theoretical model and research hypothesis. Section 2 describes the data and methodology. We report the results of our estimation models in Section 3. Sections 4 and 5 discuss theoretical and policy implications, respectively.

Literature overview and research hypothesis

Defining cross-border ownership and understanding its legitimate determinants

Assume a company Alpha, registered in any country i, is controlled by two shareholders—a natural and a legal person—based in a country j (Figure 1). We define such situation and the two shareholding ties, as ‘cross-border ownership link’. Why is this happening? Why should a company located in i-country be owned by a shareholder of j-country nationality? There can be many reasons. While most ownership links have fair and legal economic explanations, some may be driven by illicit purposes, like the need for concealing tax evasion, money laundering, and other IFFs. For example, an individual trying to hide his/her identity or the origin of his/her funds may decide to set up a holding company in country j (like company Beta in the example below), because this jurisdiction has lower corporate transparency standards and makes it more difficult for investigators or banks to identify the ultimate owner and understand where the funds come from.

Figure 1. Cross-border ownership link between countries of companies (i) and shareholders (j)

The contention of this paper is that suspicious ownership links between any i-country (location of the controlled firm) and j-country (location of the shareholder) can be detected by comparing an ideal scenario, in which cross-border ownership links are explained by legitimate determinants only, with the observed reality, in which also illicit motivations count. In other words, that all the ownership links that exceed what is predicted by legitimate drivers could suggest the existence of IFFs between countries.

The first step of this approach is therefore to understand what are the legitimate drivers explaining the ownership links between countries. We focus only on ownership links between legal persons—like limited companies, individual firms or other legal entities (e.g. the tie between company Alpha and company Beta, in the above example), since, as demonstrated by a large body of literature, legal persons are much more frequently used as veils for hiding IFF schemes (Savona and Riccardi 2017, 2018; Steinko 2012; Unger et al. 2014; van der Does de Willebois et al. 2011). A good starting point for the identification of legitimate drivers are gravity models on trade and finance. The baseline of gravity models (see Section 2 – Data and methodology for details) implies that cross-border links mirror the same rules of Newton’s gravitational theory: the closer the countries, and the bigger their size, the higher the extent of their ties. Therefore, in our case, the smaller the geographic distance between two countries, and the bigger their economies (both in terms of the real economy and the financial market), the bigger the volume of their cross-border ownership links to be expected. “Close countries” could be interpreted also in terms of cultural and social proximity: speaking the same language, and having current and former political and institutional relationships may increase trust, reduce the barriers for legal trade, and facilitate the access to valuable information between two countries (Ghemawat 2001; Sgrignoli et al. 2015)—and eventually increase the volume of corporate ownership links.

We supplement and adjust the traditional gravity model with a set of further variables which, according to the literature, could help explaining corporate ownership. First, profitability. Companies of nationality j could invest (and acquire share capital) of firms in country i in expectation of return on investment. Profitability can be interpreted also in terms of tax incentives. Investors may prefer to set up holding companies or parent companies in a country with a favourable tax rate so that they can then shift profits to the shareholders and minimise the overall fiscal pressure on the business group (Devereux et al. 2002; Zucman 2013). This would then be reflected in a lower tax rate for j-country (where the parent companies would be registered) than in i-country. Another important factor of cross-border corporate ownership is how easy it is to set up a business (World Bank 2019). The quicker and more efficient the process to set up a legitimate business in a certain j-country, the more likely it becomes that foreign investors will go there to establish a parent company.

Financial secrecy as a driver of cross-border ownership links

Although geographical, social, economic, and tax determinants are fundamental in explaining cross-border ownership links, they may not be enough. We believe that another crucial driver is the need for secrecy. Any individual willing to control a company in any i-country, but at the same time required to conceal his/her identity or the origin of his/her funds, may opt to control the firm through another legal company set up in a country j where, due to high level of financial and corporate opacity, it would be difficult to trace back his/her beneficial ownership. For this purpose, legal persons based in those j-countries with low transparency may be exploited to ultimately control companies based in i-country where transparency is higher.

There are several reasons why investors may decide to hide behind opaque corporate veils. Some may be licit—e.g. personal privacy—, but generally they are driven by illicit purposes—e.g. concealing tax evasion, money laundering, and other types of IFFs (Janský and Kokeš 2016; van der Does de Willebois et al. 2011). Financial crimes are characterised by a high degree of rationality of their actors (Gilmour, 2016; Benson & Madensesn, 2010). Following a rational choice approach, criminals who want to hide their identity would choose countries with a lower risk of detection. All money laundering techniques are aimed at layering, i.e. putting distance between the dirty proceeds and their origin and beneficial owners (Levi 2014). Among existing layering possibilities, the employment of legal firms as shell entities is a common modus operandi. In particular, when these firms are set up in countries with low corporate transparency requirements, which do not offer full disclosure of corporate owners, and do not guarantee cooperation and information exchange with foreign authorities (Ferwerda et al. 2013; Tax Justice Network 2018; van der Does de Willebois et al. 2011).

The World Bank and UNODC study on 150 cases of laundering grand corruption proceeds demonstrates the frequent involvement of certain jurisdictions as locations of shell companies: among them, the United States, British Virgin Islands (BVI), Liechtenstein, and Bahamas, all countries characterised by low corporate transparency standards (van der Does de Willebois et al. 2011). Recently, the MORE project showed the role of some EU jurisdictions, in particular Malta, Cyprus, and some Eastern European countries, as common places in which European organised crime groups (in particular Italian mafias) set up corporate vehicles for facilitating money laundering. Focusing on Spain, in most of the 367 money laundering cases judged between 1995 and 2011 and analysed by Steinko (2012), shell companies are used to conceal IFFs and are set up in a variety of jurisdictions including Andorra, Aruba, Cayman Islands, Isle of Man, and the United States (Florida, in particular). Lists of ‘secrecy jurisdictions’ emerge also from the many journalistic leaks and investigations in the last few years, the most famous being probably the Panama Papers (in 2015) and the Paradise Papers (2017). The most frequently mentioned jurisdictions in Mossack Fonseca’s files as registered seats of shell companies are Panama, BVI, Bahamas, and Seychelles. While in Paradise Papers we find Hong Kong, United Kingdom, and the United States at the top of the list.

All these (judiciary and journalistic) evidence-based reports suggest countries which are used as locations for incorporating corporate vehicles used as layers for IFFs. However, these lists are not fully representative as they may be biased depending on the type of predicate offence under analysis (e.g. grand corruption vs. organized crime), actor (e.g. tax evaders vs. mafias) and on the location of the source (e.g. Panama papers tend to overestimate the role of Caribbean countries because Mossack and Fonseca tended to provide services in this area, Steinko’s study is Hispanic-centric).

We use a different approach. Not stemming from individual cases, but inferring from aggregate data. In particular, we believe that, after controlling for legitimate determinants (the macroeconomic, social, geographical and cultural factors), ‘abnormal’ cross-border ownership links (i.e. above what is predicted by legitimate determinants) may signal illicit financial flows between any i-country (location of the firm) and j-country (location of the shareholder). And therefore we expect a positive correlation between these ‘abnormal’ links and the level of financial secrecy of the j-countries where shareholders (parent companies) are incorporated (Hypothesis H1).

Corruption as a cost in IFF schemes

If higher levels of secrecy in j-country may increase the number of foreign shareholders based in that country, what would be the role of corruption and rule of law? Ceteris paribus, are foreign shareholders more numerous from countries with high or low levels of corruption? What is the interplay between corruption and secrecy?

Even if the question is crucial in terms of policy design, the role of corruption in determining money laundering and illicit financial flows has not been investigated to any great extent, especially in empirical terms. Existing studies—mostly theoretical—do not report the same direction of causality and the literature on the relationship between corruption and IFFs is ambiguous about the sign (see Chaikin and Sharman 2009 for a review). Walker (1999) assumes that criminals do not like—excessively—corrupt countries, because corruption increases the costs of money laundering due to necessary side payments and bribes. On the other hand, Unger (2013) argues that a low level of corruption may make it difficult to find facilitators for hiding and laundering IFFs. Dreher and Schneider (2010) find empirical evidence that the relationship between shadow economy and corruption is not straightforward either: corruption reduces the shadow economy in high-income countries but increases it in low-income ones. Savona and Riccardi (2018) show that corruption is correlated to intensity in the use of cash, which in turn is a facilitator of money laundering and integration of illicit proceeds. Finally, the Basel AML Index considers corruption a risk factor of money laundering, implying that more corruption is related to more money laundering.

Based on Walker (1999), we expect that high levels of corruption in a country reduce the risk of attracting IFFs. Money laundering is a crime which is characterized by a high degree of rationality of the actors (Benson and Madensesn 2010); and money launderers, as rational agents, would move their illicit funds to those jurisdictions in which they could maximise their benefits (i.e. enjoying ill-gotten gains) and minimise their costs (i.e. the probability that proceeds are traced and predicate crimes are identified). Highly-corrupt countries with a weak rule of law would make it difficult for launderers to maximise the benefits and minimise the costs of money laundering: the extra costs imposed in terms of bribes and inefficiency would hamper the incorporation of corporate vehicles, and the weak rule of law would undermine the possibility to freely—and securely—access the proceeds once laundered. Therefore, in our setting, we test the hypothesis (H2) that ‘abnormal’ cross-border ownership links are negatively correlated with the level of corruption of j-countries (location of the shareholder): ceteris paribus, investors setting up shell companies for illicit purposes would opt for countries with a high level of secrecy but low level of corruption and stronger rule of law.

Table 1. Research hypotheses: legal and illegal determinants of cross-border ownership links

Note: Table 2 reports the variables selected to operationalise these factors and their summary statistics

Data and methodology

To test our hypothesis, we adopt a sequential-regressions strategy with 3-steps (Figure 2). The first step aims at estimating the part of cross-border ownership links which cannot be explained by legitimate reasons. At this scope, we perform a set of regression analyses, based on a gravity model, in which macroeconomic and social factors are the independent variables and the volume of cross-border ownership links is the dependent variable. Second, following the example of Cassetta and her colleagues (2014), we generate and rank the studentized Anscombe (1953) residuals emerging from the first set of models thus identifying most anomalous pair of countries corresponding to the most unpredicted (or ‘abnormal’) cross-border ownership links, which we interpret as red flags of possible IFFs.2 Third, we regress the residuals against measures of financial and corporate secrecy (to test our hypothesis H1) and of corruption and rule of law (to test H2). As a robustness check, we jointly test the explanatory power of both legitimate and illicit determinants as suggested by Chen et al. (2018).

Figure 2. Three-steps methodology

Using a gravity model to explain cross-border business ownership links

The model which we use to explain cross-border ownership links is based on a gravity model of bilateral financial flows derived from the ones theoretically and empirically investigated by authors as Eichengreen and Luengnaruemitchai (2008), Karolyi (2016), Portes, Rey, and Oh (2001). Nowadays, gravity models have become the “workhorse of applied international economics” as they allow to model characteristics of both origin and destination countries (Eichengreen and Irwin 1998). The empirical results obtained with the model have generally been judged as very good (Deardorff 1998; Ferwerda et al. 2013). The gravity model is inspired by Newton’s universal law of gravity which asserts that the attraction between two objects (F) depends on the mass of those objects (m_i and m_j), the inverse of their squared distance (r²) and the gravitational constant (G):

1. $F_{ij} = G\ \frac{m_{i}m_{j}}{{r_{ij}}^{2}}$.

By taking the logarithms of equation (1), it is possible to obtain a linear relationship which is suited to econometric analysis:

2. $\ln{F_{ij} = g + \beta_{1}\ln{m_{i} + {\beta_{2}\ln}m_{j} - \beta_{3}\ln}}r_{ij}^{2} + \varepsilon_{ij}$

For financial flows, the larger the economies and the closer the countries, the more likely it is to find ownership links among entities registered in two countries. In econometric terms:

3. $\ln{Y_{ij} = g + \beta_{1}\ln{X_{i} + {\beta_{2}\ln}X_{j} - \beta_{3}\ln}}D_{ij} + {\beta_{2}\ln}\overline{P_{ij}} + \varepsilon_{ij}$

where Y_i,j is the value of the aggregate cross-border ownership links between countries i and j. X represents the sizes of the economies of countries i and j. $D_{ij}$ denotes the distance between countries. $\overline{P_{ij}}$ represents a matrix of possible characteristics of i and j that lead shareholders to prefer a country over another one. Finally, $\varepsilon_{ij}$ are the pair-specific residuals.

We use Poisson Pseudo-Maximum Likelihood (PPML) regressions with robust standard errors clustered at i-country level to run the econometric analyses referring to step 1. PPLM regressions have emerged as the dominant estimator for empirical gravity models. PPLM allows for correctly interpreting parameters also in presence of high heteroskedasticity and it is particularly suited to deal with a large number of zeros in the dependent variable, that often characterize these models (Egger and Staub 2016; Santos Silva and Tenreyro 2006). For steps 2 and 3, we exploit studentized residuals—i.e., divided by their standard deviation—as indicators of anomaly in the ownership link, we want to make sure to clean as much as possible the error terms emerging from the regressions of step 1. Therefore, exploiting the multilevel nature of our data—i.e., level-1 pairs, level-2 countries—, we estimate the preferred model—as identified by AIC and BIC—including a destination fixed-effects, thus reducing biases due to omitted variables.

Then, we use multiple strategies to deal with the structure of the errors and the possible forms of the relationship under investigation. First, we analyse the obtained residuals through the use of a Feasible Generalised Least Squares estimator (FGLS), which are robust to heteroskedasticity and cross-sectional correlation (Fomby et al. 1984). Then, we exploit an Ordered Probit (OP) and a Probit (P) to relax any assumption of linearity in the relation between the anomalies and our variables of interest. At this scope, following Gullo and Montalbano (2018), we construct a variable that clusters the studentized residuals in three categories: <2, between 2 and 3, and >3, as 2 and 3 are common threshold values to identify outliers in the distribution of studentized residuals. We then use this variable as the dependent variable in the Ordered Probit. A dichotomous variable separating outliers—i.e. above 2—from the rest of the distribution is used in the Probit models.

Operationalisation of the ownership links

To produce our dependent variable—i.e. (the natural logarithm of) the number of shareholders of j-country of companies in i-country—we exploit business ownership information taken from the Bureau van Dijk’s ORBIS database.3 We analyse the shareholders of all companies in the database for 41 European countries. The companies in these 41 European countries (i-countries) have 28.7 million foreign shareholders (of which 9.3 million are legal persons and 781,938 foreign legal persons) from 210 countries worldwide (j-countries), whose country of origin is known (Figure 3). We aggregate the number of ownership links between all country pairs, which results in 8,610 unique pairs in which the j-country (country of the shareholder) is different from the i-country (country where the company is registered).

Figure 3. Countries of companies (i) and shareholders (j)

Note: all i-countries are simultaneously also j-countries.

Source: authors’ elaboration on Bureau van Dijk data.

The ORBIS database is the only central repository of data on business owners at the international level. Data on business owners are usually held by national business registers that cover only the firms registered in that country. Conversely, ORBIS makes it potentially possible to reproduce the entire global network of shareholders. At the same time, ORBIS provides information on both public and private companies, as well as on state-owned enterprises. For all types of companies, it provides information on all shareholders, irrespectively of their share of equity, as long as ownership information is available for that given company. For these reasons, ORBIS is used in empirical analysis in the business ownership domain (e.g. Garcia-Bernardo et al. 2017; Cobham and Janský 2017). Still, information on the shareholder’s country is not always available and its degree of availability varies among countries for a number of reasons as differences in company law, privacy rules, and the accessibility of company registries accessed by ORBIS. To account for this heterogeneity, we control for the share of available information on the nationality of business owners in each i-country.

Operationalisation of the legitimate determinants of cross-border ownership

Operationalisation of the independent variables representing the legitimate determinants of cross-border business ownership exploits open-access databases commonly used in macro-level economic and sociological studies (Table 2). In particular, the economic size of the countries considered is estimated in terms of GNI. The geographic distance of each pair of countries is operationalised by a) their physical distance weighted for the location of the population within the countries and b) by the fact of sharing a border. We rely on the stock market capitalisation and alternatively on bank deposits as measures of the size of the financial markets. The size or the value of the trades between the two countries is, instead, not included among the regressors as highly collinear with several of the other economic determinants.

We retrieve data on the nominal corporate income tax rate from KPMG (2017) and integrate them with data furnished by Deloitte (2018) to operationalise the fiscal pressure in the countries considered. The number of days required to open a business (data gathered from the WB) is used as a proxy for the bureaucratic efficiency of a country, as previously done, among the others, by Dreher and Schneider (2010). Being the analysis at the country level (and not at firm-level), we use GDP growth as a proxy of profitability. The higher the growth of country i’s economy, the more likely becomes, ceteris paribus, the attraction of foreign investments and therefore of foreign business owners (Aitken et al. 1996). Conceptually, these factors proxy the convenience of investing in i-country rather than in j-country. Therefore, they do not enter our linear equation separately for i and j, but instead in the form of the difference between the value in i and the value in j.

Finally, we operationalise social, cultural, and institutional factors. In particular, we consider if the i and the j-country are EU Member States and members of the WTO, if they were formerly part of the same country or if they have had a colonial relationship. In the expectation that cultural and social proximity influences cross-border business ownerships, we include among the regressors a) migration flows between j and i-countries as reported and b) the presence of a common language spoken by more than 9% of the population in any pair of countries.4

Operationalisation of financial secrecy, rule of law and control of corruption

We alternate the use of four different variables to measure financial secrecy. First, we exploit a dummy variable produced by Tax Justice Network (2011) which indicates whether a country can be considered a tax haven. Second, we build a variable that combines different lists of tax havens from 13 studies.5 In particular, a country could obtain a score ranging from 0 to 13; in the latter case if it is indicated as a tax haven by all 13 studies. Finally, we use the 2018 Financial Secrecy Score (FSS) estimated by Tax Justice Network (2018).6 Contrary to ‘blacklists’ of tax havens, the FSS is not a binary division between black-listed and white-listed countries; it locates countries along a secrecy spectrum ranging from countries with very high transparency (e.g. Finland) to ones with very low transparency (e.g. Vanuatu) (Cobham et al. 2015).7 Secrecy jurisdictions are often also tax havens (Gara and De Franceschis 2015); nonetheless, by controlling for the tax rate of countries i and j, as mentioned above, we isolate those links which are not driven by tax optimisation purposes but by criminal ones.

Quantifying the quality of governance and corruption is challenging and limits in cross-national and temporal comparability characterize all available measures (Kaufmann et al. 2011). Specifically, operationalising the concept of rule of law is challenging as it comprises two aspects: the existence of certain rules and how they are enforced (Kaufmann et al. 2011). The WB (2017) provides an indicator of the rule of law for 215 countries. In particular, the WB’s (2017, 1) rule of law indicator “[…] captures perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence”.

Indicators of corruption specifically focusing on business registrations and controls are not currently available; therefore, we have to rely on more general measures of corruption. Among the possible indicators of corruption, we use the Control of Corruption indicator of the WB (2017). An alternative estimate of the level of corruption in a given country is the Corruption Perceptions Index (CPI) produced by Transparency International. The correlation between the two indicators is above 0.97 for the year and the countries available in both databases (Antonaccio and Tittle 2007; Butler et al. 2007). So the use of one or the other should not affect the final estimates. We use the WB indicator because of the wider agreement among scholars on the soundness of its methodology.

Table 2. Variables operationalisation and summary statistics

Note: The reported number of observations refer to country pairs i – j, given by the combination of the 41 i-countries and 210 j-countries. Not all variables are available for all country pairs (8,610), leading to different number of observations.

^a See footnote 6 for the list of original sources.

Empirical Results

Here we describe the results of our econometric analyses. In the first set of models (Table 3), we start from a basic gravity model based on countries’ economic size and geographical distance (Model LE.1); then we add the complete set of control variables related to legal determinants of cross-border ownership (Models LE.2 to LE.7). Finally, in model LE.8, we include i-country fixed effects to control for unobserved heterogeneity.

As hypothesised, the number of foreign shareholders from j-country is positively correlated with the GNI of both i and j-countries, while it is negatively correlated with the geographic distance between the countries: the bigger the economies and the closer the countries, the higher the number of ownership links. In Model LE.2, we introduce controls for the EU and WTO memberships of both the i and j-countries. The number of foreign shareholders is, ceteris paribus, higher whenever the j-country is part of the EU. This is not surprising considering that 27 out of the 41 countries included in the sample are in the EU. The influence of WTO membership is instead weak, probably because most countries in the world are part of the organisation. Model LE.3 includes our proxy for the social, cultural, and legal proximity between j and i-countries. Countries in which part of the population speak the same language, as well as countries that have been in a colonial relationship, have a stronger connection in terms of companies’ ownership. Conversely, the size of the population migrating from shareholder-country to company-country and the fact of having been part of the same country are not correlated to the number of shareholders in this specification of the model.

Models LE.4 and LE.5 add the differential between any pair of countries in terms of procedures required to open a business, tax rates, and GDP growth (as a measure of ‘return on investment’). Models LE.6 and LE.7 include the size of the financial sector measured either by bank deposits or market capitalisation as a share of GDP, respectively. The ease of setting up a business is positively correlated with the number of foreign shareholders. Conversely, foreign shareholders are negatively correlated with the differential in the corporate tax rate, a result which apparently contradicts the extensive literature on profit-shifting (Cobham and Janský 2017; Zucman 2013). The proposed proxies for the importance of the financial sector are positively correlated when considering the j-country where the owner is located, while they are not significant when focusing on the location of the company—i.e. i-country. When controlling for the relevance of the financial sector, the correlation between the size of migrant communities and the number of foreign shareholders becomes significant, although the data considered refer to legal persons only. Finally, in model LE.8 in which we use i-country fixed effects, the size of bank deposits and the EU membership of i-countries emerge as significant predictors of foreign shareholders.

Table 3. Legitimate drivers of cross-border ownership links

Note: the table reports the standardised beta coefficients and the clusterised robust standard errors (in parentheses) of Poisson pseudo-maximum likelihood regressions of real economy, financial market, demographic, and macropolitical-related variables on the number of international shareholders from all over the world in a sample of 40 to 41 European countries. All continuous variables enter in the regression in the form of natural logarithm. The Akaike's (AIC) and the Bayesian information criteria (BIC) values provide two measures of the relative quality of the models. *, **, and ***, indicate coefficients significantly different from zero at the 95.0%, 99.0%, and 99.9% confidence level, respectively.

Models belonging to the first set present the ideal scenario in which only legal determinants of foreign investments are considered. Therefore, ownership links that appear to be abnormally above the predicted values may be interpreted instead as ‘anomalous’ links and thus possible markers of ownership structures used for IFFs. Residuals reveal these abnormalities. For this purpose, we use the residuals of LE.8, which is the model showing the best goodness of fit (AIC and BIC). Residuals from the eight models are closely correlated (Table 4).

Table 4. Residuals’ correlations

Note: the table reports the correlation between the studentized Anscombe residuals obtained in models LE.1 to LE.8 together with the significance of each correlation.

By ranking the highest residuals, a number of countries, which appear as tax havens in the literature, emerge as origins of ‘anomalous’ ownership links. When we consider the j-countries with at least 500 shareholders of firms in the dataset, Belize (17 times), the Marshall Islands (17), the Seychelles (16), Panama (10) and the Bahamas (6) appear multiple times among the top 3 anomalous connections for each i-country (Table 5). Apart from these offshore jurisdictions, Norway, Liberia, and the United States also frequently appear as j-countries in abnormal links.

We can repeat the analysis by focusing on the j-countries which account for at least the 0.1% of the foreign shareholders in the i-country. In this case, a more central role of European nationals can be detected. In particular, Norway appears 25 times as a most anomalous connection, and also Cyprus emerges. A similar picture results from Table 6 reporting the 20 j-countries with the highest average residuals. Table 7 proposes instead the 20 i-countries (location of the company) with the highest residuals.

Table 5. Top-three anomalous links (highest Studentised Anscombe Residuals) by i(company)-country

Note: for each i-country, the table reports the three most anomalous connections as identified by their studentized Anscombe residual, normalized between 0-1. Two classifications are presented. In classification A, j-countries are considered only if their total number in the database is greater than 500. In classification B, j-countries are considered only if they account for at least the 0.1% of the foreign shareholders registered in country i.

Table 6. Anomalous links: top-20 j-countries by average studentized Anscombe residual

Note: the table reports the top 20 j-countries whose links are above the model prediction as expressed by their normalized studentized Anscombe residual. LE.8 is the model of reference.

Table 7. Anomalous links: top-20 i-countries by average studentized Anscombe residual

Note: the table reports the top 20 i-countries whit most anomalous connections as identified by their studentized Anscombe residual. LE.8 with fixed effects for j-countries is the model of reference.

To investigate whether these anomalous cross-border links are correlated to secrecy, rule of law, and corruption, we run a second set of models. In these models, the residuals estimated in LE.8 are regressed on measures of financial secrecy, rule of law, and corruption. In eight out of nine econometric specifications, our indicators of financial secrecy are positively correlated to the size of the residuals (Table 8). In particular, Ordered Probit and Probit regressions always confirm a positive and significant correlation between FSS and our indicator of anomaly (IFF.3.OP and IFF.3.P). The exception is the 2018 Financial Secrecy Score, which is negatively correlated with the size of the residuals when using an FGLS strategy.

Table 8. Illicit drivers of cross-border ownership links: financial secrecy

Note: the table reports the standardised beta coefficients of Feasible Generalised Least Square (FGLS), Ordered Probit (OP), and Probit (P) regressions of variables representing tax havens on different forms of the residuals emerging from the empirical specification modelling licit determinants of transnational shareholding (i.e. LE.8). Dependent variables are: normalized residuals in FGLSs; three categories of the residuals—i.e., below 2; 2 to 3; above 3—in OPs; dichotomous division of the residuals—i.e., below 2; 2 and above—in Ps. *, **, and ***, indicate coefficients significantly different from zero at the 95.0%, 99.0%, and 99.9% confidence level, respectively.

Regressions IFF.4 to IFF.7 test the relationship between rule of law, corruption and anomalous ownership links (Table 9). The results of these additional models indicate that stronger rule of law and control of corruption in j-country (i.e. shareholder location) are positively correlated to anomalous links. The level of the rule of law and the control of corruption in i-country are not significant (IFF.5 and IFF.7). Table 10 combines both legitimate and IFF-related determinants of transnational ownership structures. The sign and significance of the main relation of interest are confirmed by these models.

Table 9. Illicit drivers of cross-border ownership links: rule of law and control of corruption

Note: the table reports the standardised beta coefficients of Feasible Generalised Least Square (FGLS), Ordered Probit (OP), and Probit (P) regressions of variables representing the perceived efficacy of the rule of law and of the control of corruption on different forms of the residuals emerging from the empirical specification modelling licit determinants of transnational shareholding (i.e. LE.8). Dependent variables are: normalized residuals in FGLSs; three categories of the residuals—i.e., below 2; 2 to 3; above 3—in OPs; dichotomous division of the residuals—i.e., below 2; 2 and above—in Ps. *, **, and ***, indicate coefficients significantly different from zero at the 95.0%, 99.0%, and 99.9% confidence level, respectively.

Table 10. Complete Models (Licit Economy and IFFs)

Note: the table reports the standardized beta coefficients and the clusterised robust standard errors (in parentheses) of Poisson pseudo-maximum likelihood regressions of the number of international shareholders in a sample of 40 European countries. In these models we use some controls for tax havens to address the potential use of companies in IFFs schemes. Models 1, 3, 5 include also indicators of the strength of the Rule of Law while Models 2, 4, 6 of Corruption Control. These variables help in understanding the nature of the countries where illicit schemes take place. All continuous variables enter in the regression in the form of natural logarithm. The AIC and the BIC values provide two measures of the relative quality of the models. *, **, and *** indicate coefficients significantly different from zero at the 95.0%, 99.0%, and 99.9% confidence level, respectively.

Discussion

The results of our analysis confirm our first hypothesis (H1) that anomalous cross-border ownership links are explained, among other factors, by the financial secrecy of the country where the shareholders are. In almost all specifications, financial secrecy is strongly correlated with our indicators of anomalous ownership links—i.e., the levels of residuals and their outliers. This result confirms a large amount of literature pointing to the opacity of the financial, banking, and corporate sectors as a key vulnerability for IFFs (Aziani 2018; Janský and Kokeš 2016; van der Does de Willebois et al. 2011).

This empirical result argues in favour of the fact that ‘rational choice’ is a strong theoretical perspective for understanding the mechanism behind IFFs, as it is for money laundering and tax fraud (Cornish and Clarke 2002; Gilmour 2016; Mehlkop and Graeff 2010). Opportunities displace money flows for IFFs toward countries where risks are lower because the financial system is more opaque.

The empirical findings highlight a strong positive relation between the control of corruption and anomalous ownership links; the stronger the corruption in a country, the lower the amount of ‘anomalous’ owners located in that country. Despite limits intrinsic to any measurement of corruption, this result confirms our second hypothesis (H2): corruption works as an impediment rather than a facilitator for IFFs. Rational investors wanting to set up shell companies to conceal illicit activities opt for jurisdictions with a higher level of secrecy but a lower level of corruption (Walker 1999).

Criminals do not have to rely on heavy corruption, as far as the instruments used to conceal IFFs are efficient. “How can a briber be sure that what he paid for will meet his expectations? Lemons are a risk to be seriously considered in this murky environment, where partners are by definition unscrupulous and no legal recourse can be sought to sanction eventual frauds, while honesty and good faith are highly appreciated virtues in bribery” (della Porta and Vannucci 2012, 18).

Finally, the correlation between the rule of law and anomalous ownership links is positive and strongly significant. Criminals prefer to direct their IFFs to more stable and peaceful countries. This further corroborates the idea that reliable institutions are a driver behind transnational illicit flows. In particular, the results presented in Table 7 show that the companies registered in Central and Eastern European countries—often former Soviet countries (i.e. Russia, Ukraine, Hungary, Czech Republic, Poland, Latvia) tend to have a higher share of anomalous ownership links. How to interpret this result? On the one hand, it could be argued that the lack of trust in financial and political institutions in these countries may foster financial outflows which, in turn, could generate a higher number of holding companies of foreign nationality (Howard 2002; Mishler and Rose 1997; Shlapentokh 2006). On the other hand, the recent political turmoils in some of these areas (e.g. Russia, Ukraine, Hungary) and the increase in power of non-state groups (Mulford 2016; Zabyelina 2019) and organised crime (Galeotti 2017; Holmes 2009) may have induced local entrepreneurs to secure their capital in foreign entities. In any case, there is evidence that corporates and financial institutions in some of these Eastern European countries have been used as conduits to launder illicit proceeds originating from former Soviet countries. See, for instance, the role of the Latvian and Estonian business sector in the Troika laundromat investigation (OCCRP 2014; Savona and Riccardi 2018), or the role of Cyprus as a destination of Russian anomalous investments (Nesvetailova 2020).

Policy implications

Our analysis may help the international cooperation aimed at identifying weak nodes in the global flows of illicit funds. In particular, detecting anomalous connections among countries with an empirical data-driven approach, can inform policymakers and help them to design new red-flags and blacklists of countries that attract IFFs.

For instance, the proposed approach is capable of highlighting the shareholder-countries with the highest number of anomalous connections. The Caribbean area plays a crucial role as a location of legal-person shareholders related to anomalous ownership links of companies registered in Europe. These results confirm previous studies in this field, such as Garcia-Bernardo et al. (2017) and the large body of evidence furnished by well-known journalistic investigations like the Panama and Paradise Papers. Regrettably, Caribbean countries do not appear in official international blacklists related to anti-money laundering or tax evasion. Among the 20 shareholder-countries with the highest average residuals (Table 6), only Bahamas is listed in the FATF AML grey list;8 only Bahamas and Panama in the EU AML blacklist of third countries;9 only Belize, Marshall Islands and Dominica in the EU blacklist of non-cooperative tax jurisdictions.10 Actually, also European countries pop up as shareholder-countries with abnormal links, but they are not included in any blacklist. Cyprus, Iceland, Luxembourg, Malta, Montenegro, and Switzerland, but also Norway, Serbia, Denmark, and Sweden rank among the first 20 shareholder-countries for which legitimate determinants fail to fully explain transnational ownership structures. In reality, these names do not surprise as they appear in previous literature related to both money laundering and organised crime investigations, especially Cyprus, Malta, and Switzerland (Gara and De Franceschis 2015).

The available data—limited to first level shareholders—do not unveil the entire ownership structure; therefore, we are not able to test if a shareholder in any shareholder-country acts as intermediate or ultimate owner. Still, we can check whether countries showing high residuals as shareholder-country (Table 6) also show high residuals as company-countries (Table 7). Countries appearing in both lists are likely to play a role as locations of intermediate owners or conduits, to use the term employed by Garcia-Bernardo et al. (2017). A beneficial owner of, say, Russian origin, in order to control a company located in Poland may use as intermediate owner a company located, say, in Cyprus. In this situation, Cyprus would appear as shareholder-country of the ownership chain, even though the ultimate owner is Russian. Bearing in mind that the samples of company and shareholder-countries are not the same, the only three jurisdictions which appear in both the lists are Cyprus, Luxembourg, and Switzerland, indicating that these three countries play some role as conduits or intermediate owners in anomalous ownership links (especially the first two).11 The role of Switzerland, in particular, warrants some further discussion. While the country frequently appears in investigations related to various forms of IFFs (Ferwerda and Reuter 2018; Zucman 2013), previous literature shows that Switzerland, rather than being the location of shell companies, plays a key role at the global level as a location for foreign bank accounts in which to store illicit funds (see eg. Van der Does de Willebois et al. 2011).

Nonetheless, none of these European countries is internationally blacklisted. Given the weakness of these official lists, due to geopolitical biases (Sharman 2009, 2012), the empirical perspective proposed by this paper could be used as an additional approach in the identification of ‘high-risk’ countries and in better guiding political pressures toward those countries actually favouring the proliferation of IFFs.

This paper is only a first step towards a better understanding of cross-border ownership links concealing IFFs. Future analysis should go beyond the first level of shareholders and map the whole network of anomalous ownership links. Moreover, the relationship between IFFs and corruption (not as a predicate offence, but as a facilitator or obstacle of IFF) requires in-depth investigation, possibly with the employment of more solid measures of corruption, also at a local level. Furthermore, it would be important to eventually disentangle the different forms and underlying crimes and activities of IFFs (see also Reuter 2017) to get a better understanding of the components of IFFs and to what extent their drivers are similar.

References

Aitken, Brian, Ann Harrison, and Robert E. Lipsey. 1996. “Wages and Foreign Ownership A Comparative Study of Mexico, Venezuela, and the United States.” Journal of International Economics, Symposium on Growth and International Trade: Empirical Studies, 40 (3): 345–71.

Alstadsæter, Annette, Niels Johannesen, and Gabriel Zucman. 2017. “Who Owns the Wealth in Tax Havens? Macro Evidence and Implications for Global Inequality.” Working Paper 23805. National Bureau of Economic Research.

Anderson, James E. 1979. “A Theoretical Foundation for the Gravity Equation.” The American Economic Review 69 (1): 106–16.

Anscombe, Frank J. 1953. ‘Sequential Estimation’. Journal of the Royal Statistical Society: Series B (Methodological) 15 (1): 1–21.

Antonaccio, Olena, and Charles R. Tittle. 2007. “A Cross-National Test of Bonger’s Theory of Criminality and Economic Conditions.” Criminology 45 (4): 925–58.

Aziani, Alberto. 2018. Illicit Financial Flows: An Innovative Approach to Estimation. International and Comparative Criminology. New York: Springer International Publishing.

Becker, Gary S. 1968. ‘Crime and Punishment: An Economic Approach’. Journal of Political Economy 76: 169–217.

Benson, Michael L., and Tamara D. Madensesn. 2010. ‘Situational Crime Prevention and White Collar Crime’. In International Handbook of White-Collar and Corporate Crime. New York: Springer.

Butler, Christopher K., Tali Gluch, and Neil J. Mitchell. 2007. “Security Forces and Sexual Violence: A Cross-National Analysis of a Principal—Agent Argument.” Journal of Peace Research 44 (6): 669–87.

Cassetta, Alessia, Claudio Pauselli, Lucia Rizzica, and Marco Tonello. 2014. “Financial Flows to Tax Havens: Determinants and Anomalies.” Quaderni Dell’antiriciclaggio 1: 1–29.

CEPII. 2017. “CEPII Databases.” CEPII. http://www.cepii.fr/cepii/en/bdd_modele/bdd.asp.

Chaikin, David, and Jason Sharman. 2009. Corruption and Money Laundering: A Symbiotic Relationship. Springer.

Chen, Wei, Paul Hribar, and Samuel Melessa. 2018. ‘Incorrect Inferences When Using Residuals as Dependent Variables’. Journal of Accounting Research 56 (3): 751–96.

Cobham, Alex, and Petr Janský. 2017. “Measuring Misalignment: The Location of US Multinationals’ Economic Activity versus the Location of Their Profits.” Development Policy Review, 48.

Cobham, Alex, Petr Janský, and Markus Meinzer. 2015. “The Financial Secrecy Index: Shedding New Light on the Geography of Secrecy.” Economic Geography 91 (3): 281–303.

Cornish, Derek B., and Ronald V. Clarke. 2002. ‘Analyzing Organized Crimes’. In Rational Choice and Criminal Behavior: Recent Research and Future Challenges, by Alex R. Piquero and Stephen G. Tibbetts. New York: Garland.

Deardorff, Alan. 1998. “Determinants of Bilateral Trade: Does Gravity Work in a Neoclassical World?” In The Regionalization of the World Economy, by Jeffrey A. Frankel, 7–32. Cambridge, MA: National Bureau of Economic Research, Inc.

della Porta, Donatella, and Alberto Vannucci. 2012. The Hidden Order of Corruption: An Institutional Approach. 1st edition. Burlington, VT: Routledge.

Deloitte. 2018. “Corporate Tax Rates 2014-2018.” KPMG. https://www2.deloitte.com/content/dam/Deloitte/global/Documents/Tax/dttl-tax-corporate-tax-rates-2014-2018.pdf.

Devereux, Michael P., Rachel Griffith, and Alexander Klemm. 2002. “Corporate Income Tax Reforms and International Tax Competition.” Economic Policy 17 (35): 449–95.

Dreher, Axel, and Friedrich Schneider. 2010. “Corruption and the Shadow Economy: An Empirical Analysis.” Public Choice 144 (1): 215–38.

Egger, Peter H., and Kevin E. Staub. 2016. ‘GLM Estimation of Trade Gravity Models with Fixed Effects’. Empirical Economics 50 (1): 137–75.

Eichengreen, Barry, and Douglas A. Irwin. 1998. “The Role of History in Bilateral Trade Flows.” In The Regionalization of the World Economy, by Jeffrey A. Frankel, 33–62. Cambridge, MA: National Bureau of Economic Research, Inc.

Eichengreen, Barry, and Pipat Luengnaruemitchai. 2008. ‘Bond Markets as Conduits for Capital Flows: How Does Asia Compare?’ International Financial Issues in the Pacific Rim: Global Imbalances, Financial Liberalization, and Exchange Rate Policy, July, 267–313.

European Commission. 2017. “Report from the Commission to the European Parliament and to the Council on the Assessment of the Risks of Money Laundering and Terrorist Financing Affecting the Internal Market and Relating to Cross-Border Situations.” Brussels: European Commission. http://ec.europa.eu/newsroom/just/item-detail.cfm?item_id=81272.

FATF. 2002. “Review to Identify Non‐Cooperative Countries or Territories: Increasing the Worldwide Effectiveness of Anti‐Money Laundering Measures.” Paris: Financial Action Tax Force on Money Laundering – FATF.

———. 2012. “International Standards on Combating Money Laundering and the Financing of Terrorism & Proliferation. The FATF Recommendations.” Paris: Financial Action Task Force - Organization for Economic Cooperation and Development.

———. 2016. “FATF Report to the G20 on Beneficial Ownership.” Paris: Financial Action Task Force.

Ferwerda, Joras, Mark Kattenberg, Han-Hsin Chang, Brigitte Unger, Loek Groot, and Jacob A. Bikker. 2013. “Gravity Models of Trade-Based Money Laundering.” Applied Economics 45 (22): 3170–82.

Ferwerda, Joras, and Peter Reuter. 2019. “Learning from Money Laundering National Risk Assessments: The Case of Italy and Switzerland.” European Journal on Criminal Policy and Research 25 (1): 5–20.

Fomby, Thomas B., Stanley R. Johnson, and R. Carter Hill. 1984. ‘Feasible Generalized Least Squares Estimation’. In Advanced Econometric Methods, edited by Thomas B. Fomby, Stanley R. Johnson, and R. Carter Hill, 147–69. New York, NY: Springer.

FSF. 2000. “Report of the Working Group on Offshore Financial Centres.” Basel: Financial Stability Forum.

Gara, Mario, and Pierpaolo De Franceschis. 2015. “I Paradisi Fiscali: Caratteristiche Operative, Evidenze Empiriche e Anomalie Finanziarie.” Quaderni Dell’antiriciclaggio 3: 28.

Garcia-Bernardo, Javier, Jan Fichtner, Frank W. Takes, and Eelke M. Heemskerk. 2017. “Uncovering Offshore Financial Centers: Conduits and Sinks in the Global Corporate Ownership Network.” Scientific Reports 7 (1): 6246.

Ghemawat, Pankaj. 2001. “Distance Still Matters. The Hard Reality of Global Expansion.” Harvard Business Review 79 (8): 137–40, 142.

Gilmour, Nicholas. 2016. ‘Understanding the Practices behind Money Laundering – A Rational Choice Interpretation’. International Journal of Law, Crime and Justice 44: 1–13.

Gravelle, Jane. 2015. “Tax Havens: International Tax Avoidance and Evasion.” Whashington DC: Congressional Research Service. https://digitalcommons.ilr.cornell.edu/key_workplace/1378.

Gullo, Valentina, and Pierluigi Montalbano. 2018. ‘Where Does “Dirty” Money Go? A Gravity Analysis’. Sapienza Working Papers (Dipartimento Di Scienze Sociali Ed Economiche) 5: 54.

Hines, James R. Jr. 2010. “Treasure Islands.” Journal of Economic Perspectives 24 (4): 103–26.

Holmes, Leslie. 2009. ‘Crime, Organised Crime and Corruption in Post-Communist Europe and the CIS’. Communist and Post-Communist Studies, The Ghosts of the Past: 20 years after the Fall of Communism in Europe, 42 (2): 265–87.

Howard, Marc Morje. 2002. “The Weakness of Postcommunist Civil Society.” Journal of Democracy 13 (1): 157–69.

Karolyi, G. Andrew. 2016. ‘The Gravity of Culture for Finance’. Journal of Corporate Finance 41 (December): 610–25.

Kaufmann D, Kraay A, Mastruzzi M 2011. “The worldwide governance indicators: methodology and analytical issues.” Hague Journal Rule Law 3 (2) :220–246

ICIJ. 2017. “Paradise Papers: Secrets of the Global Elite.” International Consortium of Investigative Journalists, 2017, Online edition. https://www.icij.org/investigations/paradise-papers/.

———. 2018. “The Panama Papers: Exposing the Rogue Offshore Finance Industry.” International Consortium of Investigative Journalists, 2018, Online edition. https://www.icij.org/investigations/panama-papers/.

IMF. 2000. “Offshore Financial Centers.” Policy Papers. Whashington DC: International Monetary Fund.

Janský, Petr, and Ondřej Kokeš. 2016. “Profit-Shifting from Czech Multinational Companies to European Tax Havens.” Applied Economics Letters 23 (16): 1130–33.

Johannesen, Niels, and Gabriel Zucman. 2014. “The End of Bank Secrecy? An Evaluation of the G20 Tax Haven Crackdown.” American Economic Journal: Economic Policy 6 (1): 65–91.

Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2011. “The Worldwide Governance Indicators: Methodology and Analytical Issues.” Hague Journal on the Rule of Law 3 (2): 220–46.

KPMG. 2017. “Corporate Tax Rates Table.” KPMG. https://home.kpmg.com/xx/en/home/services/tax/tax-tools-and-resources/tax-rates-online/corporate-tax-rates-table.html.

Levi, Michael. 2014. ‘Money Laundering’. In The Oxford Handbook of Organized Crime, edited by Letizia Paoli, 419–33. Oxford University Press.

Levin, Carl. 2007. “S. 681 (110th): Stop Tax Haven Abuse Act.” Washington, DC: US Senate.

Lowtax. 2008. “Offshore Business Review.” Lowtax. 2008. https://www.lowtax.net/offshore-business-review/.

Mehlkop, Guido, and Peter Graeff. 2010. ‘Modelling a Rational Choice Theory of Criminal Action: Subjective Expected Utilities, Norms, and Interactions’. Rationality and Society 22 (2): 189–222.

Mishler, William, and Richard Rose. 1997. ‘Trust, Distrust and Skepticism: Popular Evaluations of Civil and Political Institutions in Post-Communist Societies’. The Journal of Politics 59 (2): 418–51.

Mulford, Joshua P. 2016. ‘Non-State Actors in the Russo-Ukrainian War’. Connections 15 (2): 89–107.

Nesvetailova, Anastasia. 2020. “When Wild West Met Wild East? Understanding the Special Place of the Bahamas in Russia’s Offshore Network.” Empirical AML Research Conference Working papers (1): 26.

OCCRP. 2014. “The Russian Laundromat.” OCCRP - Organized Crime and Corruption Reporting Project, 2014. https://www.reportingproject.net/therussianlaundromat/russian-laundromat.php.

OECD. 2000. “2000 Progress Report: Towards Global Tax Co-Operation: Progress in Identifying and Eliminating Harmful Tax Practices.” Paris: OECD.

O’Regan, Philip. 2016. Financial Information Analysis. The Role of Accounting Information in Modern Society. 3rd. ed. London - New York: Routledge.

Portes, Richard, Hélène Rey, and Yonghyup Oh. 2001. ‘Information and Capital Flows: The Determinants of Transactions in Financial Assets’. European Economic Review, 15th Annual Congress of the European Economic Association, 45 (4): 783–96.

Reuter, P., 2017. Illicit financial flows and governance: The importance of disaggregation. World Bank.

Ross, Stephen, Randolph Westerfield, and Jeffrey Jaffe. 2012. Corporate Finance, 10th Edition. McGraw-Hill Education.

———. 2018. Mapping the Risk of Organised Crime Infiltration in European Businesses - Final Report of Project MORE. Università Cattolica del Sacro Cuore. MIlano. In course of publication.

Santos Silva, João M. C., and Silvana Tenreyro. 2006. ‘The Log of Gravity’. The Review of Economics and Statistics 88 (4): 641–658.

Savona, Ernesto U., and Michele Riccardi, eds. 2017. Assessing the Risk of Money Laundering in Europe - Final Report of Project IARM. Milano: Transcrime - Università Cattolica Sacro Cuore.

———. 2018. Mapping the Risk of Organised Crime Infiltration in European Businesses - Final Report of Project MORE. Università Cattolica del Sacro Cuore. Milano.

Sgrignoli, Paolo, Rodolfo Metulini, Stefano Schiavo, and Massimo Riccaboni. 2015. “The Relation Between Global Migration and Trade Networks.” Physica A: Statistical Mechanics and Its Applications 417 (January): 245–60.

Sharman, J.C. 2012. “Canaries in the Coal Mine: Tax Havens, the Decline of the West and the Rise of the Rest.” New Political Economy 17 (4): 493–513.

Sharman, J.C. 2009. “The Bark Is the Bite: International Organizations and Blacklisting.” Review of International Political Economy Vol 16, No 4 (November).

Shlapentokh, Vladimir. 2006. “Trust in Public Institutions in Russia: The Lowest in the World.” Communist and Post-Communist Studies 39 (2): 153–74.

Steinko, Armando Fernández. 2012. ‘Financial Channels of Money Laundering in Spain’. British Journal of Criminology 52 (5): 908–31.

Tax Justice Network. 2005. “Identifying Tax Havens and Offshore Finance Centres.” Chesham, UK: Tax Justice Network.

———. 2011. “Mapping Financial Secrecy: Archive 2011.” FInancial Secrecy Index. 2011. https://www.financialsecrecyindex.com/Archive2011/index.php?id=2.

———. 2018. “Financial Secrecy Index.” Financial Secrecy Index. 2018. https://www.financialsecrecyindex.com/.

UN. 2016. “United Nations Global Migration Database.” United Nations Economic and Social Affairs. 2016. https://esa.un.org/unmigration/.

Unger, Brigitte. 2013. “Money Laundering Regulation: From Al Capone to Al Qaeda.” In Research Handbook on Money Laundering, edited by Brigitte Unger and Daan van der Linde, 19–32. Cheltenham, UK: Edward Elgar Pub.

UNODC. 2017. “Building a Statistical Framework to Measure Illicit Financial Flows.” Background paper. Vienna: United Nations Office on Drugs and Crime.

van der Does de Willebois, Emile, Emily M. Halter, Robert A. Harrison, Ji Won Park, and J. C. Sharman. 2011. The Puppet Masters: How the Corrupt Use Legal Structures to Hide Stolen Assets and What to Do About It. 1 edition. Washington, DC: World Bank Publications.

Walker, John. 1999. “How Big Is Global Money Laundering?” Journal of Money Laundering Control 3 (1): 25–37.

World Bank. 2017. “World Development Indicators.” The World Bank. 2017. https://datacatalog.worldbank.org/dataset/world-development-indicators.

———. 2019. “Doing Business.” Text/HTML. World Bank. 2019. http://www.doingbusiness.org/en/doingbusiness.

Zabyelina, Yuliya. 2019. ‘Vigilante Justice and Informal Policing in Post-Euromaidan Ukraine’. Post-Soviet Affairs 35 (4): 277–92.

Zoromé, Ahmed. 2007. “Concept of Offshore Financial Centers: In Search of an Operational Definition.” IMF Working Paper 07 (87): 1–34.

Zucman, Gabriel. 2013. “The Missing Wealth of Nations: Are Europe and the U.S. Net Debtors or Net Creditors?” The Quarterly Journal of Economics 128 (3): 1321–64.